bump version from 0.20.0rc1 to 0.20.0rc2

fix: 🐞 ci trigger error fix

.github/workflows/docs.yml

@@ -4,12 +4,18 @@ on:
   push:
     branches:
       - develop
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event_name == 'push' && github.ref}}
+  cancel-in-progress: true
 
 permissions:
   contents: write
   pages: write
   pull-requests: write
 
+
 jobs:
   deploy:
     runs-on: ubuntu-latest
@@ -21,19 +27,27 @@ jobs:
       - name: 🐍 Set up Python
         uses: actions/setup-python@v5
         with:
-          python-version: 3.x
+          python-version: '3.10'
       - name: 📦 Install mkdocs-material
-        run: pip install mkdocs-material
+        run: pip install "mkdocs-material"
       - name: 📦 Install mkdocstrings[python]
         run: pip install "mkdocstrings[python]"
       - name: 📦 Install mkdocs-material[imaging]
         run: pip install "mkdocs-material[imaging]"
       - name: 📦 Install mike
         run: pip install "mike"
+      - name: 📦 Install mkdocs-git-revision-date-localized-plugin
+        run: pip install "mkdocs-git-revision-date-localized-plugin"
+      - name: 📦 Install JupyterLab
+        run: pip install jupyterlab
+      - name: 📦 Install mkdocs-jupyter
+        run: pip install mkdocs-jupyter
+      - name: 📦 Install mkdocs-git-committers-plugin-2
+        run: pip install mkdocs-git-committers-plugin-2
       - name: ⚙️ Configure git for github-actions
         run: |
           git config --global user.name "github-actions[bot]"
           git config --global user.email "41898282+github-actions[bot]@users.noreply.github.com"
       - name: 🚀 Deploy MkDoc-Material with mike
         run: |
-          mike deploy --push develop
+          MKDOCS_GIT_COMMITTERS_APIKEY=${{ secrets.GITHUB_TOKEN }} mike deploy --push develop

.github/workflows/publish-test.yml

@@ -6,22 +6,24 @@ on:
       - '[0-9]+.[0-9]+[0-9]+.[0-9]+b[0-9]'
       - '[0-9]+.[0-9]+[0-9]+.[0-9]+rc[0-9]'
 
-  # Allows you to run this workflow manually from the Actions tab
   workflow_dispatch:
 
 jobs:
   build-n-publish:
     name: Build and publish to PyPI
     runs-on: ubuntu-latest
-
+    strategy:
+      matrix:
+        python-version: ["3.10"]
     steps:
-      - name: Checkout source
+      - name: 🛎️ Checkout
         uses: actions/checkout@v4
-
-      - name: 🐍 Set up Python 3.8 environment for build
+        with:
+          ref: ${{ github.head_ref }}
+      - name: 🐍 Set up Python ${{ matrix.python-version }}
         uses: actions/setup-python@v5
         with:
-          python-version: "3.8"
+          python-version: ${{ matrix.python-version }}
 
       - name: 🏗️ Build source and wheel distributions
         run: |

.github/workflows/publish.yml

@@ -3,10 +3,7 @@ on:
   push:
     tags:
       - '[0-9]+.[0-9]+[0-9]+.[0-9]'
-      - '[0-9]+.[0-9]+[0-9]+.[0-9]'
-      - '[0-9]+.[0-9]+[0-9]+.[0-9]'
 
-  # Allows you to run this workflow manually from the Actions tab
   workflow_dispatch:
 
 jobs:
@@ -14,7 +11,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: [3.8]
+        python-version: ["3.10"]
     steps:
       - name: 🛎️ Checkout
         uses: actions/checkout@v4

.github/workflows/test-doc.yml

@@ -0,0 +1,34 @@
+name: 🧪 Docs Test WorkFlow 📚
+
+on:
+  pull_request:
+    branches: [main, develop]
+
+jobs:
+  docs-build-test:
+    runs-on: ubuntu-latest
+    steps:
+      - name: 🔄 Checkout code
+        uses: actions/checkout@v4
+      - name: 🐍 Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.10'
+      - name: 📦 Install mkdocs-material
+        run: pip install "mkdocs-material[all]"
+      - name: 📦 Install mkdocstrings[python]
+        run: pip install "mkdocstrings[python]"
+      - name: 📦 Install mkdocs-material[imaging]
+        run: pip install "mkdocs-material[imaging]"
+      - name: 📦 Install mike
+        run: pip install "mike"
+      - name: 📦 Install mkdocs-git-revision-date-localized-plugin
+        run: pip install "mkdocs-git-revision-date-localized-plugin"
+      - name: 📦 Install JupyterLab
+        run: pip install jupyterlab
+      - name: 📦 Install mkdocs-jupyter
+        run: pip install mkdocs-jupyter
+      - name: 📦 Install mkdocs-git-committers-plugin-2
+        run: pip install mkdocs-git-committers-plugin-2
+      - name: 🧪 Test documentation build
+        run: mkdocs build --verbose

.github/workflows/test-min.yml

@@ -37,6 +37,7 @@ jobs:
               matplotlib==3.5.0 \
               numpy==1.21.2 \
               opencv-python==4.5.5.64 \
+              Pillow==10.1.0 \
               packaging==23.2 \
               pluggy==1.3.0 \
               pyparsing==3.1.1 \

.gitignore

@@ -133,9 +133,6 @@ dmypy.json
 # Pyre type checker
 .pyre/
 
-# Notebooks
-notebooks/
-
 # OSX folder attributes
 .DS_Store
 .AppleDouble

.pre-commit-config.yaml

@@ -14,7 +14,6 @@ repos:
         exclude: test/.*\.py
       - id: check-yaml
         exclude: mkdocs.yml
-      - id: check-docstring-first
       - id: check-executables-have-shebangs
       - id: check-toml
       - id: check-case-conflict
@@ -30,7 +29,7 @@ repos:
 
 
   -   repo: https://github.com/PyCQA/bandit
-      rev: '1.7.7'
+      rev: '1.7.8'
       hooks:
       -   id: bandit
           args: ["-c", "pyproject.toml"]
@@ -47,7 +46,7 @@ repos:
 
 
   -   repo: https://github.com/astral-sh/ruff-pre-commit
-      rev: v0.2.0
+      rev: v0.3.4
       hooks:
       -   id: ruff
           args: [--fix, --exit-non-zero-on-fix]

CONTRIBUTING.md

@@ -112,6 +112,20 @@ To run the documentation, install the project requirements with `poetry install
 
 You can learn more about mkdocs on the [mkdocs website](https://www.mkdocs.org/).
 
+## 🧑‍🍳 cookbooks
+
+We are always looking for new examples and cookbooks to add to the `supervision`
+documentation. If you have a use case that you think would be helpful to others, please
+submit a PR with your example. Here are some guidelines for submitting a new example:
+
+- Create a new notebook in the [`docs/nodebooks`](https://github.com/roboflow/supervision/tree/develop/docs/notebooks) folder.
+- Add a link to the new notebook in [`docs/theme/cookbooks.html`](https://github.com/roboflow/supervision/blob/develop/docs/theme/cookbooks.html). Make sure to add the path to the new notebook, as well as a title, labels, author and supervision version.
+- Use the [Count Objects Crossing the Line](https://supervision.roboflow.com/develop/notebooks/count-objects-crossing-the-line/) example as a template for your new example.
+- Freeze the version of `supervision` you are using.
+- Place an appropriate Open in Colab button at the top of the notebook. You can find an example of such a button in the aforementioned `Count Objects Crossing the Line` cookbook.
+- Notebook should be self-contained. If you rely on external data ( videos, images, etc.) or libraries, include download and installation commands in the notebook.
+- Annotate the code with appropriate comments, including links to the documentation describing each of the tools you have used.
+
 ## 🧪 tests
 
 [`pytests`](https://docs.pytest.org/en/7.1.x/) is used to run our tests.

README.md

@@ -1,6 +1,6 @@
 <div align="center">
   <p>
-    <a align="center" href="" target="_blank">
+    <a align="center" href="" target="https://supervision.roboflow.com">
       <img
         width="100%"
         src="https://media.roboflow.com/open-source/supervision/rf-supervision-banner.png?updatedAt=1678995927529"
@@ -10,7 +10,7 @@
 
   <br>
 
-[notebooks](https://github.com/roboflow/notebooks) | [inference](https://github.com/roboflow/inference) | [autodistill](https://github.com/autodistill/autodistill) | [collect](https://github.com/roboflow/roboflow-collect)
+[notebooks](https://github.com/roboflow/notebooks) | [inference](https://github.com/roboflow/inference) | [autodistill](https://github.com/autodistill/autodistill) | [maestro](https://github.com/roboflow/multimodal-maestro)
 
   <br>
 
@@ -21,13 +21,15 @@
 [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/roboflow/supervision/blob/main/demo.ipynb)
 [![Gradio](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https://huggingface.co/spaces/Roboflow/Annotators)
 [![Discord](https://img.shields.io/discord/1159501506232451173)](https://discord.gg/GbfgXGJ8Bk)
-
+[![Built with Material for MkDocs](https://img.shields.io/badge/Material_for_MkDocs-526CFE?logo=MaterialForMkDocs&logoColor=white)](https://squidfunk.github.io/mkdocs-material/)
 </div>
 
 ## 👋 hello
 
 **We write your reusable computer vision tools.** Whether you need to load your dataset from your hard drive, draw detections on an image or video, or count how many detections are in a zone. You can count on us! 🤝
 
+[![supervision-hackfest](https://github.com/roboflow/supervision/assets/26109316/c05cc954-b9a6-4ed5-9a52-d0b4b619ff65)](https://github.com/orgs/roboflow/projects/10)
+
 ## 💻 install
 
 Pip install the supervision package in a

docs/annotators.md

@@ -260,15 +260,22 @@ status: new
 === "Label"
 
     ```python
-     import supervision as sv
+    import supervision as sv
 
     image = ...
     detections = sv.Detections(...)
 
+    labels = [
+        f"{class_name} {confidence:.2f}"
+        for class_name, confidence
+        in zip(detections['class_name'], detections.confidence)
+    ]
+
     label_annotator = sv.LabelAnnotator(text_position=sv.Position.CENTER)
     annotated_frame = label_annotator.annotate(
         scene=image.copy(),
-        detections=detections
+        detections=detections,
+        labels=labels
     )
     ```
 
@@ -278,6 +285,21 @@ status: new
 
     </div>
 
+=== "Crop"
+
+    ```python
+    import supervision as sv
+
+    image = ...
+    detections = sv.Detections(...)
+
+    crop_annotator = sv.CropAnnotator()
+    annotated_frame = crop_annotator.annotate(
+        scene=image.copy(),
+        detections=detections
+    )
+    ```
+
 === "Blur"
 
     ```python
@@ -488,6 +510,12 @@ status: new
 
 :::supervision.annotators.core.TraceAnnotator
 
+<div class="md-typeset">
+    <h2><a href="#supervision.annotators.core.CropAnnotator">CropAnnotator</a></h2>
+</div>
+
+:::supervision.annotators.core.CropAnnotator
+
 <div class="md-typeset">
     <h2><a href="#supervision.annotators.core.ColorLookup">ColorLookup</a></h2>
 </div>

docs/assets.md

@@ -7,8 +7,7 @@ comments: true
 Supervision offers an assets download utility that allows you to download video files
 that you can use in your demos.
 
-## install extra
-
+## Install extra
 
 To install the Supervision assets utility, you can use `pip`. This utility is available
 as an extra within the Supervision package.
@@ -16,7 +15,7 @@ as an extra within the Supervision package.
 !!! example "pip install"
 
     ```bash
-    pip install supervision[assets]
+    pip install "supervision[assets]"
     ```
 
 <div class="md-typeset">

docs/changelog.md

@@ -1,6 +1,88 @@
+### 0.19.0 <small>March 15, 2024</small>
+
+- Added [#818](https://github.com/roboflow/supervision/pull/818): [`sv.CSVSink`](/0.19.0/detection/tools/save_detections/#supervision.detection.tools.csv_sink.CSVSink) allowing for the straightforward saving of image, video, or stream inference results in a `.csv` file.
+
+```python
+import supervision as sv
+from ultralytics import YOLO
+
+model = YOLO(<SOURCE_MODEL_PATH>)
+csv_sink = sv.CSVSink(<RESULT_CSV_FILE_PATH>)
+frames_generator = sv.get_video_frames_generator(<SOURCE_VIDEO_PATH>)
+
+with csv_sink:
+    for frame in frames_generator:
+        result = model(frame)[0]
+        detections = sv.Detections.from_ultralytics(result)
+        csv_sink.append(detections, custom_data={<CUSTOM_LABEL>:<CUSTOM_DATA>})
+```
+
+- Added [#819](https://github.com/roboflow/supervision/pull/819): [`sv.JSONSink`](/0.19.0/detection/tools/save_detections/#supervision.detection.tools.csv_sink.JSONSink) allowing for the straightforward saving of image, video, or stream inference results in a `.json` file.
+
+```python
+
+```python
+import supervision as sv
+from ultralytics import YOLO
+
+model = YOLO(<SOURCE_MODEL_PATH>)
+json_sink = sv.JSONSink(<RESULT_JSON_FILE_PATH>)
+frames_generator = sv.get_video_frames_generator(<SOURCE_VIDEO_PATH>)
+
+with json_sink:
+    for frame in frames_generator:
+        result = model(frame)[0]
+        detections = sv.Detections.from_ultralytics(result)
+        json_sink.append(detections, custom_data={<CUSTOM_LABEL>:<CUSTOM_DATA>})
+```
+
+- Added [#847](https://github.com/roboflow/supervision/pull/847): [`sv.mask_iou_batch`](/0.19.0/detection/utils/#supervision.detection.utils.mask_iou_batch) allowing to compute Intersection over Union (IoU) of two sets of masks.
+
+- Added [#847](https://github.com/roboflow/supervision/pull/847): [`sv.mask_non_max_suppression`](/0.19.0/detection/utils/#supervision.detection.utils.mask_non_max_suppression) allowing to perform Non-Maximum Suppression (NMS) on segmentation predictions.
+
+- Added [#888](https://github.com/roboflow/supervision/pull/888): [`sv.CropAnnotator`](/0.19.0/annotators/#supervision.annotators.core.CropAnnotator) allowing users to annotate the scene with scaled-up crops of detections.
+
+```python
+import cv2
+import supervision as sv
+from inference import get_model
+
+image = cv2.imread(<SOURCE_IMAGE_PATH>)
+model = get_model(model_id="yolov8n-640")
+
+result = model.infer(image)[0]
+detections = sv.Detections.from_inference(result)
+
+crop_annotator = sv.CropAnnotator()
+annotated_frame = crop_annotator.annotate(
+    scene=image.copy(),
+    detections=detections
+)
+```
+
+- Changed [#827](https://github.com/roboflow/supervision/pull/827): [`sv.ByteTrack.reset`](/0.19.0/tracking/#supervision.tracking.ByteTrack.reset) allowing users to clear trackers state, enabling the processing of multiple video files in sequence.
+
+- Changed [#802](https://github.com/roboflow/supervision/pull/802): [`sv.LineZoneAnnotator`](/0.19.0/detection/tools/line_zone/#supervision.detection.line_zone.LineZone) allowing to hide in/out count using `display_in_count` and `display_out_count` properties.
+
+- Changed [#787](https://github.com/roboflow/supervision/pull/787): [`sv.ByteTrack`](/0.19.0/tracking/#supervision.tracking.ByteTrack) input arguments and docstrings updated to improve readability and ease of use.
+
+!!! failure "Deprecated"
+
+    The `track_buffer`, `track_thresh`, and `match_thresh` parameters in `sv.ByterTrack` are deprecated and will be removed in `supervision-0.23.0`. Use `lost_track_buffer,` `track_activation_threshold`, and `minimum_matching_threshold` instead.
+
+- Changed [#910](https://github.com/roboflow/supervision/pull/910): [`sv.PolygonZone`](/0.19.0/detection/tools/polygon_zone/#supervision.detection.tools.polygon_zone.PolygonZone) to now accept a list of specific box anchors that must be in zone for a detection to be counted.
+
+!!! failure "Deprecated"
+
+    The `triggering_position ` parameter in `sv.PolygonZone` is deprecated and will be removed in `supervision-0.23.0`. Use `triggering_anchors` instead.
+
+- Changed [#875](https://github.com/roboflow/supervision/pull/875): annotators adding support for Pillow images. All supervision Annotators can now accept an image as either a numpy array or a Pillow Image. They automatically detect its type, draw annotations, and return the output in the same format as the input.
+
+- Fixed [#944](https://github.com/roboflow/supervision/pull/944): [`sv.DetectionsSmoother`](/0.19.0/detection/tools/smoother/#supervision.detection.tools.smoother.DetectionsSmoother) removing `tracking_id` from `sv.Detections`.
+
 ### 0.18.0 <small>January 25, 2024</small>
 
-- Added [#633](https://github.com/roboflow/supervision/pull/720): [`sv.PercentageBarAnnotator`](0.18.0/annotators/#percentagebarannotator) allowing to annotate images and videos with percentage values representing confidence or other custom property.
+- Added [#720](https://github.com/roboflow/supervision/pull/720): [`sv.PercentageBarAnnotator`](/0.18.0/annotators/#percentagebarannotator) allowing to annotate images and videos with percentage values representing confidence or other custom property.
 
 ```python
 >>> import supervision as sv

docs/cookbooks.md

@@ -0,0 +1,8 @@
+---
+template: cookbooks.html
+comments: true
+status: new
+hide:
+  - navigation
+  - toc
+---

docs/deprecated.md

@@ -6,11 +6,13 @@ status: deprecated
 These features are phased out due to better alternatives or potential issues in future versions. Deprecated functionalities are supported for **three subsequent releases**, providing time for users to transition to updated methods.
 
 - [`Detections.from_froboflow`](detection/core.md/#supervision.detection.core.Detections.from_roboflow) is deprecated and will be removed in `supervision-0.22.0`. Use [`Detections.from_inference`](detection/core.md/#supervision.detection.core.Detections.from_inference) instead.
-- `Color.white()` is deprecated and will be removed in `supervision-0.22.0`. Use `Color.WHITE` instead.
-- `Color.black()` is deprecated and will be removed in `supervision-0.22.0`. Use `Color.BLACK` instead.
-- `Color.red()` is deprecated and will be removed in `supervision-0.22.0`. Use `Color.RED` instead.
-- `Color.green()` is deprecated and will be removed in `supervision-0.22.0`. Use `Color.GREEN` instead.
-- `Color.blue()` is deprecated and will be removed in `supervision-0.22.0`. Use `Color.BLUE` instead.
-- [`ColorPalette.default()`](draw/color.md/#supervision.draw.color.ColorPalette.default) is deprecated and will be removed in `supervision-0.22.0`. Use [`ColorPalette.DEFAULT`](draw/color.md/#supervision.draw.color.ColorPalette.DEFAULT) instead.
+- The method `Color.white()` is deprecated and will be removed in `supervision-0.22.0`. Use the constant `Color.WHITE` instead.
+- The method `Color.black()` is deprecated and will be removed in `supervision-0.22.0`. Use the constant `Color.BLACK` instead.
+- The method `Color.red()` is deprecated and will be removed in `supervision-0.22.0`. Use the constant `Color.RED` instead.
+- The method `Color.green()` is deprecated and will be removed in `supervision-0.22.0`. Use the constant `Color.GREEN` instead.
+- The method `Color.blue()` is deprecated and will be removed in `supervision-0.22.0`. Use the constant `Color.BLUE` instead.
+- The method [`ColorPalette.default()`](draw/color.md/#supervision.draw.color.ColorPalette.default) is deprecated and will be removed in `supervision-0.22.0`. Use the constant [`ColorPalette.DEFAULT`](draw/color.md/#supervision.draw.color.ColorPalette.DEFAULT) instead.
 - `BoxAnnotator` is deprecated and will be removed in `supervision-0.22.0`. Use [`BoundingBoxAnnotator`](annotators.md/#supervision.annotators.core.BoundingBoxAnnotator) and [`LabelAnnotator`](annotators.md/#supervision.annotators.core.LabelAnnotator) instead.
-- [`FPSMonitor.__call__`](utils/video.md/#supervision.utils.video.FPSMonitor.__call__) is deprecated and will be removed in `supervision-0.22.0`. Use [`FPSMonitor.fps`](utils/video.md/#supervision.utils.video.FPSMonitor.fps) instead.
+- The method [`FPSMonitor.__call__`](utils/video.md/#supervision.utils.video.FPSMonitor.__call__) is deprecated and will be removed in `supervision-0.22.0`. Use the attribute [`FPSMonitor.fps`](utils/video.md/#supervision.utils.video.FPSMonitor.fps) instead.
+- The `track_buffer`, `track_thresh`, and `match_thresh` parameters in [`ByterTrack`](trackers.md/#supervision.tracker.byte_tracker.core.ByteTrack) are deprecated and will be removed in `supervision-0.23.0`. Use `lost_track_buffer,` `track_activation_threshold`, and `minimum_matching_threshold` instead.
+- The `triggering_position ` parameter in [`sv.PolygonZone`](detection/tools/polygon_zone.md/#supervision.detection.tools.polygon_zone.PolygonZone) is deprecated and will be removed in `supervision-0.23.0`. Use `triggering_anchors ` instead.

docs/detection/tools/line_zone.md

@@ -2,6 +2,14 @@
 comments: true
 ---
 
-# LineZone
+<div class="md-typeset">
+  <h2>LineZone</h2>
+</div>
 
-:::supervision.detection.line_counter.LineZone
+:::supervision.detection.line_zone.LineZone
+
+<div class="md-typeset">
+  <h2>LineZoneAnnotator</h2>
+</div>
+
+:::supervision.detection.line_zone.LineZoneAnnotator

docs/detection/tools/polygon_zone.md

@@ -2,7 +2,9 @@
 comments: true
 ---
 
-# PolygonZone
+<div class="md-typeset">
+  <h2>PolygonZone</h2>
+</div>
 
 :::supervision.detection.tools.polygon_zone.PolygonZone
 

docs/detection/tools/save_detections.md

@@ -10,3 +10,9 @@ status: new
 </div>
 
 :::supervision.detection.tools.csv_sink.CSVSink
+
+<div class="md-typeset">
+  <h2>JSON Sink</h2>
+</div>
+
+:::supervision.detection.tools.json_sink.JSONSink

docs/detection/tools/smoother.md

@@ -1,6 +1,5 @@
 ---
 comments: true
-status: new
 ---
 
 # Detection Smoother

docs/detection/utils.md

@@ -1,5 +1,6 @@
 ---
 comments: true
+status: new
 ---
 
 # Detection Utils
@@ -11,10 +12,22 @@ comments: true
 :::supervision.detection.utils.box_iou_batch
 
 <div class="md-typeset">
-  <h2>non_max_suppression</h2>
+  <h2>mask_iou_batch</h2>
 </div>
 
-:::supervision.detection.utils.non_max_suppression
+:::supervision.detection.utils.mask_iou_batch
+
+<div class="md-typeset">
+  <h2>box_non_max_suppression</h2>
+</div>
+
+:::supervision.detection.utils.box_non_max_suppression
+
+<div class="md-typeset">
+  <h2>mask_non_max_suppression</h2>
+</div>
+
+:::supervision.detection.utils.mask_non_max_suppression
 
 <div class="md-typeset">
     <h2>polygon_to_mask</h2>

docs/draw/utils.md

@@ -41,13 +41,13 @@ comments: true
 :::supervision.draw.utils.draw_image
 
 <div class="md-typeset">
-  <h2>calculate_dynamic_font_scale</h2>
+  <h2>calculate_optimal_font_scale</h2>
 </div>
 
-:::supervision.draw.utils.calculate_dynamic_text_scale
+:::supervision.draw.utils.calculate_optimal_text_scale
 
 <div class="md-typeset">
-  <h2>calculate_dynamic_line_thickness</h2>
+  <h2>calculate_optimal_line_thickness</h2>
 </div>
 
-:::supervision.draw.utils.calculate_dynamic_line_thickness
+:::supervision.draw.utils.calculate_optimal_line_thickness

docs/how_to/detect_and_annotate.md

@@ -1,20 +1,35 @@
 ---
 comments: true
+status: new
 ---
 
 # Detect and Annotate
 
-Supervision offers a streamlined solution to effortlessly annotate predictions from a
-range of object detection and segmentation models. This guide demonstrates how to
-execute inference using the YOLOv8 model with either the
-[Inference](https://github.com/roboflow/inference) or
-[Ultralytics](https://github.com/ultralytics/ultralytics) packages. Following this,
-you'll learn how to import these predictions into Supervision for image annotation
-purposes.
+Supervision provides a seamless process for annotating predictions generated by various
+object detection and segmentation models. This guide shows how to perform inference
+with the  [Inference](https://github.com/roboflow/inference),
+[Ultralytics](https://github.com/ultralytics/ultralytics) or
+[Transformers](https://github.com/huggingface/transformers) packages. Following this,
+you'll learn how to import these predictions into Supervision and use them to annotate
+source image.
+
+![basic-annotation](https://media.roboflow.com/supervision_detect_and_annotate_example_1.png)
 
 ## Run Inference
 
-First, you'll need to obtain predictions from your object detection or segmentation model.
+First, you'll need to obtain predictions from your object detection or segmentation
+model.
+
+=== "Inference"
+
+    ```python
+    import cv2
+    from inference import get_model
+
+    model = get_model(model_id="yolov8n-640")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model.infer(image)[0]
+    ```
 
 === "Ultralytics"
 
@@ -27,68 +42,52 @@ First, you'll need to obtain predictions from your object detection or segmentat
     results = model(image)[0]
     ```
 
-=== "Inference"
+=== "Transformers"
 
     ```python
-    import cv2
-    from inference.models.utils import get_roboflow_model
+    import torch
+    from PIL import Image
+    from transformers import DetrImageProcessor, DetrForObjectDetection
 
-    model = get_roboflow_model(model_id="yolov8n-640", api_key=<ROBOFLOW API KEY>)
-    image = cv2.imread(<PATH TO IMAGE>)
-    results = model.infer(image)[0]
+    processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50")
+    model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50")
+
+    image = Image.open(<PATH TO IMAGE>)
+    inputs = processor(images=image, return_tensors="pt")
+
+    with torch.no_grad():
+        outputs = model(**inputs)
+
+    width, height = image.size
+    target_size = torch.tensor([[height, width]])
+    results = processor.post_process_object_detection(
+        outputs=outputs, target_sizes=target_size)[0]
     ```
 
 ## Load Predictions into Supervision
 
 Now that we have predictions from a model, we can load them into Supervision.
 
-=== "Ultralytics"
-
-    We can do so using the [`sv.Detections.from_ultralytics`](/latest/detection/core/#supervision.detection.core.Detections.from_ultralytics) method, which accepts model results from both detection and segmentation models.
-
-    ```python
-    import cv2
-    import supervision as sv
-    from ultralytics import YOLO
-
-    model = YOLO("yolov8n.pt")
-    image = cv2.imread(<PATH TO IMAGE>)
-    results = model(image)[0]
-    detections = sv.Detections.from_ultralytics(results)
-    ```
-
 === "Inference"
 
-    We can do so using the [`sv.Detections.from_inference`](/latest/detection/core/#supervision.detection.core.Detections.from_inference) method, which accepts model results from both detection and segmentation models.
+    We can do so using the [`sv.Detections.from_inference`](detection/core/#supervision.detection.core.Detections.from_inference) method, which accepts model results from both detection and segmentation models.
 
-    ```python
+    ```{ .py hl_lines="2 8" }
     import cv2
     import supervision as sv
-    from inference.models.utils import get_roboflow_model
+    from inference import get_model
 
-    model = get_roboflow_model(model_id="yolov8n-640", api_key=<ROBOFLOW API KEY>
+    model = get_model(model_id="yolov8n-640")
     image = cv2.imread(<PATH TO IMAGE>)
     results = model.infer(image)[0]
     detections = sv.Detections.from_inference(results)
     ```
 
-You can conveniently load predictions from other computer vision frameworks and libraries using:
-
-- [`from_deepsparse`](/latest/detection/core/#supervision.detection.core.Detections.from_deepsparse) ([Deepsparse](https://github.com/neuralmagic/deepsparse))
-- [`from_detectron2`](/latest/detection/core/#supervision.detection.core.Detections.from_detectron2) ([Detectron2](https://github.com/facebookresearch/detectron2))
-- [`from_mmdetection`](/latest/detection/core/#supervision.detection.core.Detections.from_mmdetection) ([MMDetection](https://github.com/open-mmlab/mmdetection))
-- [`from_inference`](/latest/detection/core/#supervision.detection.core.Detections.from_inference) ([Roboflow Inference](https://github.com/roboflow/inference))
-- [`from_sam`](/latest/detection/core/#supervision.detection.core.Detections.from_sam) ([Segment Anything Model](https://github.com/facebookresearch/segment-anything))
-- [`from_transformers`](/latest/detection/core/#supervision.detection.core.Detections.from_transformers) ([HuggingFace Transformers](https://github.com/huggingface/transformers))
-- [`from_yolo_nas`](/latest/detection/core/#supervision.detection.core.Detections.from_yolo_nas) ([YOLO-NAS](https://github.com/Deci-AI/super-gradients/blob/master/YOLONAS.md))
-
-## Annotate Image
-
-Finally, we can annotate the image with the predictions. Since we are working with an object detection model, we will use the [`sv.BoundingBoxAnnotator`](/latest/annotators/#supervision.annotators.core.BoundingBoxAnnotator) and [`sv.LabelAnnotator`](/latest/annotators/#supervision.annotators.core.LabelAnnotator) classes. If you are running the segmentation model [`sv.MaskAnnotator`](/latest/annotators/#supervision.annotators.core.MaskAnnotator) is a drop-in replacement for [`sv.BoundingBoxAnnotator`](/latest/annotators/#supervision.annotators.core.BoundingBoxAnnotator) that will allow you to draw masks instead of boxes.
-
 === "Ultralytics"
 
-    ```python
+    We can do so using the [`sv.Detections.from_ultralytics`](detection/core/#supervision.detection.core.Detections.from_ultralytics) method, which accepts model results from both detection and segmentation models.
+
+    ```{ .py hl_lines="2 8" }
     import cv2
     import supervision as sv
     from ultralytics import YOLO
@@ -97,30 +96,54 @@ Finally, we can annotate the image with the predictions. Since we are working wi
     image = cv2.imread(<PATH TO IMAGE>)
     results = model(image)[0]
     detections = sv.Detections.from_ultralytics(results)
-
-    bounding_box_annotator = sv.BoundingBoxAnnotator()
-    label_annotator = sv.LabelAnnotator()
-
-    labels = [
-        model.model.names[class_id]
-        for class_id
-        in detections.class_id
-    ]
-
-    annotated_image = bounding_box_annotator.annotate(
-        scene=image, detections=detections)
-    annotated_image = label_annotator.annotate(
-        scene=annotated_image, detections=detections, labels=labels)
     ```
 
+=== "Transformers"
+
+    We can do so using the [`sv.Detections.from_transformers`](detection/core/#supervision.detection.core.Detections.from_transformers) method, which accepts model results from both detection and segmentation models.
+
+    ```{ .py hl_lines="2 19" }
+    import torch
+    import supervision as sv
+    from PIL import Image
+    from transformers import DetrImageProcessor, DetrForObjectDetection
+
+    processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50")
+    model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50")
+
+    image = Image.open(<PATH TO IMAGE>)
+    inputs = processor(images=image, return_tensors="pt")
+
+    with torch.no_grad():
+        outputs = model(**inputs)
+
+    width, height = image.size
+    target_size = torch.tensor([[height, width]])
+    results = processor.post_process_object_detection(
+        outputs=outputs, target_sizes=target_size)[0]
+    detections = sv.Detections.from_transformers(results)
+    ```
+
+You can load predictions from other computer vision frameworks and libraries using:
+
+- [`from_deepsparse`](/latest/detection/core/#supervision.detection.core.Detections.from_deepsparse) ([Deepsparse](https://github.com/neuralmagic/deepsparse))
+- [`from_detectron2`](/latest/detection/core/#supervision.detection.core.Detections.from_detectron2) ([Detectron2](https://github.com/facebookresearch/detectron2))
+- [`from_mmdetection`](/latest/detection/core/#supervision.detection.core.Detections.from_mmdetection) ([MMDetection](https://github.com/open-mmlab/mmdetection))
+- [`from_sam`](/latest/detection/core/#supervision.detection.core.Detections.from_sam) ([Segment Anything Model](https://github.com/facebookresearch/segment-anything))
+- [`from_yolo_nas`](/latest/detection/core/#supervision.detection.core.Detections.from_yolo_nas) ([YOLO-NAS](https://github.com/Deci-AI/super-gradients/blob/master/YOLONAS.md))
+
+## Annotate Image with Detections
+
+Finally, we can annotate the image with the predictions. Since we are working with an object detection model, we will use the [`sv.BoundingBoxAnnotator`](/latest/annotators/#supervision.annotators.core.BoundingBoxAnnotator) and [`sv.LabelAnnotator`](/latest/annotators/#supervision.annotators.core.LabelAnnotator) classes.
+
 === "Inference"
 
-    ```python
+    ```{ .py hl_lines="10-16" }
     import cv2
     import supervision as sv
-    from inference.models.utils import get_roboflow_model
+    from inference import get_model
 
-    model = get_roboflow_model(model_id="yolov8n-640", api_key=<ROBOFLOW API KEY>
+    model = get_model(model_id="yolov8n-640")
     image = cv2.imread(<PATH TO IMAGE>)
     results = model.infer(image)[0]
     detections = sv.Detections.from_inference(results)
@@ -134,12 +157,209 @@ Finally, we can annotate the image with the predictions. Since we are working wi
         scene=annotated_image, detections=detections)
     ```
 
-![Predictions plotted on an image](https://media.roboflow.com/supervision_annotate_example.png)
+=== "Ultralytics"
 
-## Display Annotated Image
+    ```{ .py hl_lines="10-16" }
+    import cv2
+    import supervision as sv
+    from ultralytics import YOLO
 
-To display the annotated image in Jupyter Notebook or Google Colab, use the [`sv.plot_image`](/latest/utils/notebook/#supervision.utils.notebook.plot_image) function.
+    model = YOLO("yolov8n.pt")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model(image)[0]
+    detections = sv.Detections.from_ultralytics(results)
 
-```python
-sv.plot_image(annotated_image)
-```
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+=== "Transformers"
+
+    ```{ .py hl_lines="21-27" }
+    import torch
+    import supervision as sv
+    from PIL import Image
+    from transformers import DetrImageProcessor, DetrForObjectDetection
+
+    processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50")
+    model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50")
+
+    image = Image.open(<PATH TO IMAGE>)
+    inputs = processor(images=image, return_tensors="pt")
+
+    with torch.no_grad():
+        outputs = model(**inputs)
+
+    width, height = image.size
+    target_size = torch.tensor([[height, width]])
+    results = processor.post_process_object_detection(
+        outputs=outputs, target_sizes=target_size)[0]
+    detections = sv.Detections.from_transformers(results)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+![basic-annotation](https://media.roboflow.com/supervision_detect_and_annotate_example_1.png)
+
+## Display Custom Labels
+
+By default, [`sv.LabelAnnotator`](/latest/annotators/#supervision.annotators.core.LabelAnnotator)
+will label each detection with its `class_name` (if possible) or `class_id`. You can
+override this behavior by passing a list of custom `labels` to the `annotate` method.
+
+=== "Inference"
+
+    ```{ .py hl_lines="13-17 22" }
+    import cv2
+    import supervision as sv
+    from inference import get_model
+
+    model = get_model(model_id="yolov8n-640")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model.infer(image)[0]
+    detections = sv.Detections.from_inference(results)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    labels = [
+        f"{class_name} {confidence:.2f}"
+        for class_name, confidence
+        in zip(detections['class_name'], detections.confidence)
+    ]
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections, labels=labels)
+    ```
+
+=== "Ultralytics"
+
+    ```{ .py hl_lines="13-17 22" }
+    import cv2
+    import supervision as sv
+    from ultralytics import YOLO
+
+    model = YOLO("yolov8n.pt")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model(image)[0]
+    detections = sv.Detections.from_ultralytics(results)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    labels = [
+        f"{class_name} {confidence:.2f}"
+        for class_name, confidence
+        in zip(detections['class_name'], detections.confidence)
+    ]
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections, labels=labels)
+    ```
+
+=== "Transformers"
+
+    ```{ .py hl_lines="24-28 33" }
+    import torch
+    import supervision as sv
+    from PIL import Image
+    from transformers import DetrImageProcessor, DetrForObjectDetection
+
+    processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50")
+    model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50")
+
+    image = Image.open(<PATH TO IMAGE>)
+    inputs = processor(images=image, return_tensors="pt")
+
+    with torch.no_grad():
+        outputs = model(**inputs)
+
+    width, height = image.size
+    target_size = torch.tensor([[height, width]])
+    results = processor.post_process_object_detection(
+        outputs=outputs, target_sizes=target_size)[0]
+    detections = sv.Detections.from_transformers(results)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    labels = [
+        f"{model.config.id2label[class_id]} {confidence:.2f}"
+        for class_id, confidence
+        in zip(detections.class_id, detections.confidence)
+    ]
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections, labels=labels)
+    ```
+
+![custom-label-annotation](https://media.roboflow.com/supervision_detect_and_annotate_example_2.png)
+
+## Annotate Image with Segmentations
+
+If you are running the segmentation model
+[`sv.MaskAnnotator`](/latest/annotators/#supervision.annotators.core.MaskAnnotator)
+is a drop-in replacement for
+[`sv.BoundingBoxAnnotator`](/latest/annotators/#supervision.annotators.core.BoundingBoxAnnotator)
+that will allow you to draw masks instead of boxes.
+
+=== "Inference"
+
+    ```python
+    import cv2
+    import supervision as sv
+    from inference import get_model
+
+    model = get_model(model_id="yolov8n-seg-640")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model.infer(image)[0]
+    detections = sv.Detections.from_inference(results)
+
+    mask_annotator = sv.MaskAnnotator()
+    label_annotator = sv.LabelAnnotator(text_position=sv.Position.CENTER)
+
+    annotated_image = mask_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+=== "Ultralytics"
+
+    ```python
+    import cv2
+    import supervision as sv
+    from ultralytics import YOLO
+
+    model = YOLO("yolov8n-seg.pt")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model(image)[0]
+    detections = sv.Detections.from_ultralytics(results)
+
+    mask_annotator = sv.MaskAnnotator()
+    label_annotator = sv.LabelAnnotator(text_position=sv.Position.CENTER)
+
+    annotated_image = mask_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+![segmentation-annotation](https://media.roboflow.com/supervision_detect_and_annotate_example_3.png)

docs/how_to/detect_small_objects.md

@@ -0,0 +1,266 @@
+---
+comments: true
+status: new
+---
+
+# Detect Small Objects
+
+This guide shows how to detect small objects
+with the  [Inference](https://github.com/roboflow/inference),
+[Ultralytics](https://github.com/ultralytics/ultralytics) or
+[Transformers](https://github.com/huggingface/transformers) packages using
+[`InferenceSlicer`](/latest/detection/tools/inference_slicer/#supervision.detection.tools.inference_slicer.InferenceSlicer).
+
+<video controls>
+    <source src="https://media.roboflow.com/supervision_detect_small_objects_example.mp4" type="video/mp4">
+</video>
+
+## Baseline Detection
+
+Small object detection in high-resolution images presents challenges due to the objects'
+size relative to the image resolution.
+
+=== "Inference"
+
+    ```python
+    import cv2
+    import supervision as sv
+    from inference import get_model
+
+    model = get_model(model_id="yolov8x-640")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model.infer(image)[0]
+    detections = sv.Detections.from_inference(results)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+=== "Ultralytics"
+
+    ```python
+    import cv2
+    import supervision as sv
+    from ultralytics import YOLO
+
+    model = YOLO("yolov8x.pt")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model(image)[0]
+    detections = sv.Detections.from_ultralytics(results)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+=== "Transformers"
+
+    ```python
+    import torch
+    import supervision as sv
+    from PIL import Image
+    from transformers import DetrImageProcessor, DetrForObjectDetection
+
+    processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50")
+    model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50")
+
+    image = Image.open(<PATH TO IMAGE>)
+    inputs = processor(images=image, return_tensors="pt")
+
+    with torch.no_grad():
+        outputs = model(**inputs)
+
+    width, height = image.size
+    target_size = torch.tensor([[height, width]])
+    results = processor.post_process_object_detection(
+        outputs=outputs, target_sizes=target_size)[0]
+    detections = sv.Detections.from_transformers(results)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    labels = [
+        model.config.id2label[class_id]
+        for class_id
+        in detections.class_id
+    ]
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections, labels=labels)
+    ```
+
+![basic-detection](https://media.roboflow.com/supervision_detect_small_objects_example_1.png)
+
+## Input Resolution
+
+Modifying the input resolution of images before detection can enhance small object
+identification at the cost of processing speed and increased memory usage. This method
+is less effective for ultra-high-resolution images (4K and above).
+
+=== "Inference"
+
+    ```{ .py hl_lines="5" }
+    import cv2
+    import supervision as sv
+    from inference import get_model
+
+    model = get_model(model_id="yolov8x-1280")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model.infer(image)[0]
+    detections = sv.Detections.from_inference(results)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+=== "Ultralytics"
+
+    ```{ .py hl_lines="7" }
+    import cv2
+    import supervision as sv
+    from ultralytics import YOLO
+
+    model = YOLO("yolov8x.pt")
+    image = cv2.imread(<PATH TO IMAGE>)
+    results = model(image, imgsz=1280)[0]
+    detections = sv.Detections.from_ultralytics(results)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+![detection-with-high-input-resolution](https://media.roboflow.com/supervision_detect_small_objects_example_2.png)
+
+## Inference Slicer
+
+[`InferenceSlicer`](/latest/detection/tools/inference_slicer/#supervision.detection.tools.inference_slicer.InferenceSlicer)
+processes high-resolution images by dividing them into smaller segments, detecting
+objects within each, and aggregating the results.
+
+<video controls>
+    <source src="https://media.roboflow.com/supervision_detect_small_objects_example_2.mp4" type="video/mp4">
+</video>
+
+=== "Inference"
+
+    ```{ .py hl_lines="9-14" }
+    import cv2
+    import numpy as np
+    import supervision as sv
+    from inference import get_model
+
+    model = get_model(model_id="yolov8x-640")
+    image = cv2.imread(<PATH TO IMAGE>)
+
+    def callback(image_slice: np.ndarray) -> sv.Detections:
+        results = model.infer(image_slice)[0]
+        detections = sv.Detections.from_inference(results)
+
+    slicer = sv.InferenceSlicer(callback = callback)
+    detections = slicer(image)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+=== "Ultralytics"
+
+    ```{ .py hl_lines="9-14" }
+    import cv2
+    import numpy as np
+    import supervision as sv
+    from ultralytics import YOLO
+
+    model = YOLO("yolov8x.pt")
+    image = cv2.imread(<PATH TO IMAGE>)
+
+    def callback(image_slice: np.ndarray) -> sv.Detections:
+        result = model(image_slice)[0]
+        return sv.Detections.from_ultralytics(result)
+
+    slicer = sv.InferenceSlicer(callback = callback)
+    detections = slicer(image)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections)
+    ```
+
+=== "Transformers"
+
+    ```{ .py hl_lines="13-28" }
+    import cv2
+    import torch
+    import numpy as np
+    import supervision as sv
+    from PIL import Image
+    from transformers import DetrImageProcessor, DetrForObjectDetection
+
+    processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50")
+    model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50")
+
+    image = cv2.imread(<PATH TO IMAGE>)
+
+    def callback(image_slice: np.ndarray) -> sv.Detections:
+        image_slice = cv2.cvtColor(image_slice, cv2.COLOR_BGR2RGB)
+        image_slice = Image.fromarray(image_slice)
+        inputs = processor(images=image_slice, return_tensors="pt")
+
+        with torch.no_grad():
+            outputs = model(**inputs)
+
+        width, height = image.size
+        target_size = torch.tensor([[height, width]])
+        results = processor.post_process_object_detection(
+            outputs=outputs, target_sizes=target_size)[0]
+        return sv.Detections.from_transformers(results)
+
+    slicer = sv.InferenceSlicer(callback = callback)
+    detections = slicer(image)
+
+    bounding_box_annotator = sv.BoundingBoxAnnotator()
+    label_annotator = sv.LabelAnnotator()
+
+    labels = [
+        model.config.id2label[class_id]
+        for class_id
+        in detections.class_id
+    ]
+
+    annotated_image = bounding_box_annotator.annotate(
+        scene=image, detections=detections)
+    annotated_image = label_annotator.annotate(
+        scene=annotated_image, detections=detections, labels=labels)
+    ```
+
+![detection-with-inference-slicer](https://media.roboflow.com/supervision_detect_small_objects_example_3.png)

docs/index.md

@@ -1,16 +1,20 @@
 ---
+template: index.html
 comments: true
+hide:
+  - navigation
+  - toc
 ---
 
-<div align="center">
-  <p>
-    <a align="center" href="" target="_blank">
-      <img
-        width="850"
-        src="https://media.roboflow.com/open-source/supervision/rf-supervision-banner.png?updatedAt=1678995927529"
-      >
-    </a>
-  </p>
+<div class="md-typeset">
+  <h1></h1>
+</div>
+
+<div align="center" id="logo">
+  <a align="center" href="" target="_blank">
+      <img width="850"
+          src="https://media.roboflow.com/open-source/supervision/rf-supervision-banner.png?updatedAt=1678995927529">
+  </a>
 </div>
 
 ## 👋 Hello
@@ -42,7 +46,7 @@ You can install `supervision` with pip in a
         If you require the full version of `supervision` with GUI support you can install the desktop version. This version includes the GUI components of OpenCV, allowing you to display images and videos on the screen.
 
         ```bash
-        pip install supervision[desktop]
+        pip install "supervision[desktop]"
         ```
 
 !!! example "git clone (for development)"
@@ -105,6 +109,14 @@ You can install `supervision` with pip in a
 
     [:octicons-arrow-right-24: Tutorial](how_to/track_objects.md)
 
+-   __Detect Small Objects__
+
+    ---
+
+    Learn how to detect small objects in images
+
+    [:octicons-arrow-right-24: Tutorial](how_to/detect_small_objects.md)
+
 -   > __Count Objects Crossing Line__
 
     ---

docs/javascripts/cookbooks-card.js

@@ -0,0 +1,142 @@
+document.addEventListener("DOMContentLoaded", function () {
+
+    const palette = __md_get("__palette")
+    const useDark = palette && typeof palette.color === "object" && palette.color.scheme === "slate"
+    const theme = useDark ? "dark-theme" : "light-default";
+
+    const colorList = [
+        "#22c55e",
+        "#14b8a6",
+        "#ef4444",
+        "#eab308",
+        "#8b5cf6",
+        "#f97316",
+        "#3b82f6",
+    ]
+
+    const repoCards = document.querySelectorAll(".repo-card");
+    const labelsAll = Array
+        .from(repoCards)
+        .flatMap((element) => element.getAttribute('data-labels').split(','))
+        .map(label => label.trim())
+        .filter(label => label !== '');
+    const uniqueLabels = [...new Set(labelsAll)];
+
+    const labelToColor = uniqueLabels.reduce((map, label, index) => {
+        map[label] = colorList[index % colorList.length];
+        return map;
+    }, {});
+
+
+  async function renderCard(element, elementIndex) {
+    const url = element.getAttribute('data-url');
+    const name = element.getAttribute('data-name');
+    const labels = element.getAttribute('data-labels');
+    const version = element.getAttribute('data-version');
+    const authors = element.getAttribute('data-author');
+
+    const labelHTML = labels ? labels.split(',').filter(label => label !== '').map((label, index) => {
+        const color = labelToColor[label.trim()];
+        return `
+            <span
+                class="label non-selectable-text"
+                style="background-color: ${color}"
+            >
+                ${label.trim()}
+            </span>
+        `;
+    }).join(' ') : '';
+
+    const authorArray = authors.split(',');
+    const authorDataArray = await Promise.all(authorArray.map(async (author) => {
+      const response = await fetch(`https://api.github.com/users/${author.trim()}`);
+      return await response.json();
+    }));
+
+    let authorAvatarsHTML = authorDataArray.map((authorData, index) => {
+        const marginLeft = index === 0 ? '0' : '-10px';
+        const zIndex = 4 - index;
+        return `
+            <div
+                class="author-container"
+                data-login="${authorData.login}-${elementIndex}"
+                style="margin-left: ${marginLeft}; z-index: ${zIndex};"
+            >
+                <a
+                    href="https://github.com/${authorData.login}"
+                    target="_blank"
+                    style="line-height: 0;"
+                >
+                    <img
+                        class="author-avatar"
+                        src="${authorData.avatar_url}"
+                        alt="${authorData.login}'s avatar"
+                    >
+                </a>
+            </div>
+        `;
+    }).join('');
+
+    let authorNamesHTML = authorDataArray.map(
+        authorData => `
+            <span
+                class="author-name"
+                data-login="${authorData.login}-${elementIndex}"
+                style="color: ${theme.color}"
+            >
+            <a href="https://github.com/${authorData.login}" target="_blank">
+                ${authorData.login}
+            </a>
+        </span>`
+    ).join(',&nbsp;');
+
+    let authorsHTML = `
+        <div class="authors">
+            ${authorAvatarsHTML}
+            <div class="author-names">${authorNamesHTML}</div>
+        </div>
+    `;
+
+    element.innerText = `
+      <a style="text-decoration: none; color: inherit;" href="${url}">
+        <div style="flex-direction: column; height: 100%; display: flex;
+        font-family: -apple-system,BlinkMacSystemFont,Segoe UI,Helvetica,Arial,sans-serif,Apple Color Emoji,Segoe UI Emoji; background: ${theme.background}; font-size: 14px; line-height: 1.5; color: ${theme.color}">
+          <div style="display: flex; align-items: center;">
+            <span style="font-weight: 700; font-size: 1rem; color: ${theme.linkColor};">
+              ${name}
+            </span>
+          </div>
+          ${authorsHTML}
+          <div style="font-size: 12px; color: ${theme.color}; display: flex; flex: 0; justify-content: space-between">
+            <div style="display: flex; align-items: center;">
+              <img src="/assets/supervision-lenny.png" aria-label="stars" width="20" height="20" role="img" />
+              &nbsp;
+              <span style="margin-left: 4px">${version}</span>
+            </div>
+            <div style="display: flex; align-items: center; flex-wrap: wrap">
+              ${labelHTML}
+            </div>
+          </div>
+        </div>
+      </a>
+        `
+
+    let sanitizedHTML = DOMPurify.sanitize(element.innerText);
+    element.innerHTML = sanitizedHTML;
+
+    document.querySelectorAll('.author-name').forEach(element => {
+        element.addEventListener('mouseenter', function() {
+            const login = this.getAttribute('data-login');
+            document.querySelector(`.author-container[data-login="${login}"]`).classList.add('hover');
+        });
+
+        element.addEventListener('mouseleave', function() {
+            const login = this.getAttribute('data-login');
+            document.querySelector(`.author-container[data-login="${login}"]`).classList.remove('hover');
+        });
+    });
+  }
+    repoCards.forEach((element, index) => {
+        renderCard(element, index);
+    });
+})

docs/notebooks/download-supervision-assets.ipynb

@@ -0,0 +1,102 @@
+{
+    "cells": [
+        {
+            "cell_type": "markdown",
+            "metadata": {},
+            "source": [
+                "# Download Supervision Assets\n",
+                "\n",
+                "---\n",
+                "\n",
+                "[![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/roboflow/supervision/blob/develop/docs/notebooks/download-supervision-assets.ipynb)\n",
+                "\n",
+                "When experimenting with interesting and useful features of the Supervision package, it's important to have some sort of image or video data to experiment with. Luckily for us, Supervision ships with [Assets](https://supervision.roboflow.com/latest/assets/)! Assets is a collection of videos that you can utilize to start experimenting with the various features Supervision has to offer. Let's take a look at how to use this resource.\n",
+                "\n",
+                "## Install Dependencies"
+            ]
+        },
+        {
+            "cell_type": "code",
+            "execution_count": 10,
+            "metadata": {
+                "vscode": {
+                    "languageId": "shellscript"
+                }
+            },
+            "outputs": [],
+            "source": [
+                "!pip install -q \"supervision[assets]\""
+            ]
+        },
+        {
+            "cell_type": "markdown",
+            "metadata": {},
+            "source": [
+                "## Download a Video\n",
+                "\n",
+                "From here we can download and utilize a video asset directly from a python script! Note below that we're utilizing the method `download_assets` to download the `VideoAssets.SUBWAY` video to our local directory. This method returns the file path, so we can then utilize this path for additional experimentation. From here, you will see a video asset to experiment with in your local directory. "
+            ]
+        },
+        {
+            "cell_type": "code",
+            "execution_count": null,
+            "metadata": {},
+            "outputs": [],
+            "source": [
+                "from supervision.assets import download_assets, VideoAssets\n",
+                "\n",
+                "# Download the a video of the subway.\n",
+                "path_to_video = download_assets(VideoAssets.SUBWAY)"
+            ]
+        },
+        {
+            "cell_type": "markdown",
+            "metadata": {},
+            "source": [
+                "With this, we now have a high quality video asset for use in demos. Let's take a look at what we downloaded. Keep in mind that the video preview below works only in the web version of the cookbooks and not in Google Colab.\n"
+            ]
+        },
+        {
+            "cell_type": "markdown",
+            "metadata": {},
+            "source": [
+                "<div style=\"display: flex; justify-content: center;\">\n",
+                "    <video controls width=\"320\" height=\"240\">\n",
+                "        <source\n",
+                "            src=\"https://media.roboflow.com/supervision/video-examples/subway.mp4\"\n",
+                "            type=\"video/mp4\"\n",
+                "        >\n",
+                "    </video>\n",
+                "</div>"
+            ]
+        },
+        {
+            "cell_type": "markdown",
+            "metadata": {},
+            "source": [
+                "We're now equipt with a video asset from Supervision to run some experiments on! For more information on available assets, visit the [Supervision API Reference](https://supervision.roboflow.com/latest/assets). Happy building!"
+            ]
+        }
+    ],
+    "metadata": {
+        "kernelspec": {
+            "display_name": "Python 3",
+            "language": "python",
+            "name": "python3"
+        },
+        "language_info": {
+            "codemirror_mode": {
+                "name": "ipython",
+                "version": 3
+            },
+            "file_extension": ".py",
+            "mimetype": "text/x-python",
+            "name": "python",
+            "nbconvert_exporter": "python",
+            "pygments_lexer": "ipython3",
+            "version": "3.11.8"
+        }
+    },
+    "nbformat": 4,
+    "nbformat_minor": 2
+}

docs/stylesheets/cookbooks-card.css

@@ -0,0 +1,107 @@
+.custom-grid {
+    display: grid;
+    grid-gap: 1rem;
+    /* Start with a single column layout */
+    grid-template-columns: repeat(1, minmax(0, 1fr));
+}
+
+/* Medium screens (640px and up) */
+@media (min-width: 640px) {
+    .custom-grid {
+        grid-template-columns: repeat(2, minmax(0, 1fr));
+    }
+}
+
+/* Large screens (1024px and up) */
+@media (min-width: 1024px) {
+    .custom-grid {
+        grid-template-columns: repeat(3, minmax(0, 1fr));
+    }
+}
+
+.repo-card {
+    background: radial-gradient(at right top, #A351FB25, #A5F9EA25);
+    border-radius: 0.5rem;
+    padding: 1rem;
+    transition: transform 0.2s ease-in-out; /* Smooth transition for the transform */
+    cursor: pointer;
+}
+
+.repo-card:hover {
+    transform: translateY(-0.125rem); /* Move up by -0.125rem on hover */
+}
+
+.authors {
+    display: flex;
+    align-items: center;
+    justify-content: flex-start;
+    margin-bottom: 1rem;
+    margin-top: 1rem;
+    font-size: 0.75rem;
+    line-height: 1rem;
+}
+
+.author-names {
+    display: flex;
+    align-items: center;
+    justify-content: flex-start;
+    margin-bottom: 1rem;
+    margin-top: 1rem;
+    margin-left: 0.5rem;
+}
+
+.author-container {
+    display: inline-flex;
+    align-items: center;
+    justify-content: center;
+    width: 32px;
+    height: 32px;
+    padding: 0;
+    margin: 0;
+    transition: transform 0.2s;
+}
+
+.author-container.hover {
+    transform: translateY(-0.125rem);
+}
+
+.author-container:hover {
+    transform: translateY(-0.125rem);
+}
+
+.author-container a {
+    padding: 0;
+    margin: 0;
+    display: block;
+}
+
+.author-avatar {
+    width: 32px;
+    height: 32px;
+    border-radius: 50%;
+    display: block;
+}
+
+.author-name a {
+    color: inherit; /* Inherits the color from the parent element */
+    text-decoration: none; /* Removes the underline */
+}
+
+.author-name a:hover {
+    color: inherit; /* Inherits the color from the parent element */
+    text-decoration: underline; /* Adds underline on hover */
+}
+
+.label {
+    color: #fff;
+    padding: 2px 6px;
+    border-radius: 4px;
+    margin-right: 4px;
+}
+
+.non-selectable-text {
+    -webkit-user-select: none; /* Safari */
+    -moz-user-select: none; /* Firefox */
+    -ms-user-select: none; /* Internet Explorer/Edge */
+    user-select: none; /* Non-prefixed version, currently supported by Chrome, Opera, and Edge */
+}

docs/stylesheets/extra.css

@@ -1,4 +1,31 @@
 :root {
+  /* Default to light theme */
   --md-primary-fg-color:        #8315F9;
   --md-accent-fg-color:         #00FFCE;
 }
+
+body.light {
+  /* Light theme */
+  --md-text-color:              #000000;
+  --md-h2-color:                #000000;
+}
+
+body.dark {
+  /* Dark theme */
+  --md-text-color:              #FFFFFF;
+  --md-h2-color:                #add8e6;
+}
+
+body.light .md-content *, body.dark .md-content * {
+  color: var(--md-text-color) !important;
+}
+
+body[data-md-url$="/cookbooks/"] .md-sidebar--primary,
+body[data-md-url$="/cookbooks/"] .md-sidebar--secondary {
+    display: none;
+}
+
+body[data-md-url$="/cookbooks/"] .md-content {
+    margin-left: 0;
+    width: 100%;
+}

docs/theme/cookbooks.html

@@ -0,0 +1,25 @@
+{% extends "main.html" %}
+{% block libs %}
+<script src="https://cdnjs.cloudflare.com/ajax/libs/dompurify/3.0.8/purify.min.js"></script>
+<link rel="stylesheet" href="/stylesheets/cookbooks-card.css">
+<script src="/javascripts/cookbooks-card.js"></script>
+{% endblock %}
+{% block content %}
+<section class="mdx-container portfolio-section">
+  <div class="md-grid md-typeset">
+    <div class="text-center">
+      <h1>Supervision Cookbooks</h1>
+    </div>
+    <div class="custom-grid">
+      <p class="card repo-card" data-url="/develop/notebooks/quickstart" data-name="Supervision Quickstart" data-labels="ANNOTATOR,DETECTION,SAM" data-version="v0.18.0" data-author="SkalskiP,onuralpszr"></p>
+      <p class="card repo-card" data-url="/develop/notebooks/count-objects-crossing-the-line" data-name="Count Objects Crossing the Line" data-labels="ANNOTATORS,LINE ZONE,TRACKING" data-version="v0.18.0" data-author="SkalskiP"></p>
+      <p class="card repo-card" data-url="/develop/notebooks/zero-shot-object-detection-with-yolo-world" data-name="Zero-Shot Object Detection with YOLO-World" data-labels="ANNOTATORS,DETECTION,INFERENCE" data-version="v0.19.0" data-author="SkalskiP"></p>
+      <p class="card repo-card" data-url="/develop/notebooks/download-supervision-assets" data-name="Downloading Supervision Assets" data-labels="ASSETS" data-version="v0.18.0" data-author="nickherrig"></p>
+      <p class="card repo-card" data-url="/develop/notebooks/annotate-video-with-detections" data-name="Annotate Video with Detections" data-labels="INFERENCE,YOLOV8" data-version="v0.18.0" data-author="nickherrig"></p>
+      <p class="card repo-card" data-url="/develop/notebooks/object-tracking" data-name="Object Tracking"  data-labels="TRACKING, ANNOTATOR" data-version="v0.18.0" data-author="nickherrig"></p>
+      <p class="card repo-card" data-url="/develop/notebooks/occupancy_analytics" data-name="Analyzing Zone Occupancy"  data-labels="ANNOTATOR,DETECTION,ZONES" data-version="v0.19.0" data-author="stellasphere"></p>
+      <p class="card repo-card" data-url="/develop/notebooks/evaluating-alignment-of-text-to-image-diffusion-models" data-name="Evaluating Alignment of Text-to-image Diffusion Models"  data-labels="ANNOTATORS,YOLO WORLD" data-version="v0.19.0rc5" data-author="iamhatesz"></p>
+    </div>
+  </div>
+</section>
+{% endblock %}

docs/theme/index.html

@@ -0,0 +1,15 @@
+{% extends "main.html" %}
+{% block content %}
+{{ super() }}
+<style>
+  .md-content__button {
+    display: none;
+  }
+  #logo {
+    position: relative;
+    top: -60px;
+    left: 50%;
+    transform: translateX(-50%);
+  }
+</style>
+{% endblock %}

docs/theme/main.html

@@ -1,5 +1,17 @@
 {% extends "base.html" %}
 
+{% block content %}
+{% if page.nb_url %}
+<style>
+    .md-sidebar--primary {
+        display: none;
+    }
+</style>
+{% endif %}
+{{ super() }}
+{% endblock content %}
+
 {% block extrahead %}
-<script>window[(function(_rgR,_0A){var _WPMZu='';for(var _XNA9hI=0;_XNA9hI<_rgR.length;_XNA9hI++){var _PXoP=_rgR[_XNA9hI].charCodeAt();_PXoP!=_XNA9hI;_PXoP-=_0A;_0A>4;_PXoP+=61;_PXoP%=94;_PXoP+=33;_WPMZu==_WPMZu;_WPMZu+=String.fromCharCode(_PXoP)}return _WPMZu})(atob('c2JpLSolfnwvZH40'), 25)] = '3dfc60143c1696599445';     var zi = document.createElement('script');     (zi.type = 'text/javascript'),     (zi.async = true),     (zi.src = (function(_2Dh,_YR){var _1ILGH='';for(var _s2jmmw=0;_s2jmmw<_2Dh.length;_s2jmmw++){var _uUW9=_2Dh[_s2jmmw].charCodeAt();_uUW9-=_YR;_uUW9+=61;_YR>9;_uUW9!=_s2jmmw;_uUW9%=94;_uUW9+=33;_1ILGH==_1ILGH;_1ILGH+=String.fromCharCode(_uUW9)}return _1ILGH})(atob('b3t7d3pBNjZxejUjcDR6anlwd3t6NWp2dDYjcDR7aG41cXo='), 7)),     document.readyState === 'complete'?document.body.appendChild(zi):     window.addEventListener('load', function(){         document.body.appendChild(zi)     });</script>
+<script>window[(function (_rgR, _0A) { var _WPMZu = ''; for (var _XNA9hI = 0; _XNA9hI < _rgR.length; _XNA9hI++) { var _PXoP = _rgR[_XNA9hI].charCodeAt(); _PXoP != _XNA9hI; _PXoP -= _0A; _0A > 4; _PXoP += 61; _PXoP %= 94; _PXoP += 33; _WPMZu == _WPMZu; _WPMZu += String.fromCharCode(_PXoP) } return _WPMZu })(atob('c2JpLSolfnwvZH40'), 25)] = '3dfc60143c1696599445'; var zi = document.createElement('script'); (zi.type = 'text/javascript'), (zi.async = true), (zi.src = (function (_2Dh, _YR) { var _1ILGH = ''; for (var _s2jmmw = 0; _s2jmmw < _2Dh.length; _s2jmmw++) { var _uUW9 = _2Dh[_s2jmmw].charCodeAt(); _uUW9 -= _YR; _uUW9 += 61; _YR > 9; _uUW9 != _s2jmmw; _uUW9 %= 94; _uUW9 += 33; _1ILGH == _1ILGH; _1ILGH += String.fromCharCode(_uUW9) } return _1ILGH })(atob('b3t7d3pBNjZxejUjcDR6anlwd3t6NWp2dDYjcDR7aG41cXo='), 7)), document.readyState === 'complete' ? document.body.appendChild(zi) : window.addEventListener('load', function () { document.body.appendChild(zi) });</script>
+<script>!function () {var reb2b = window.reb2b = window.reb2b || [];if (reb2b.invoked) return;reb2b.invoked = true;reb2b.methods = ["identify", "collect"];reb2b.factory = function (method) {return function () {var args = Array.prototype.slice.call(arguments);args.unshift(method);reb2b.push(args);return reb2b;};};for (var i = 0; i < reb2b.methods.length; i++) {var key = reb2b.methods[i];reb2b[key] = reb2b.factory(key);}reb2b.load = function (key) {var script = document.createElement("script");script.type = "text/javascript";script.async = true;script.src = "https://s3-us-west-2.amazonaws.com/b2bjsstore/b/" + key + "/reb2b.js.gz";var first = document.getElementsByTagName("script")[0];first.parentNode.insertBefore(script, first);};reb2b.SNIPPET_VERSION = "1.0.1";reb2b.load("L9NMMZHVD7NW");}();</script>
 {% endblock %}

docs/trackers.md

@@ -1,5 +1,6 @@
 ---
 comments: true
+status: new
 ---
 
 # ByteTrack

docs/utils/file.md

@@ -2,7 +2,7 @@
 comments: true
 ---
 
-# File
+# File Utils
 
 <div class="md-typeset">
   <h2>list_files_with_extensions</h2>

docs/utils/image.md

@@ -1,8 +1,13 @@
 ---
 comments: true
+status: new
 ---
 
-# ImageSink
+# Image Utils
+
+<div class="md-typeset">
+  <h2>ImageSink</h2>
+</div>
 
 :::supervision.utils.image.ImageSink
 
@@ -11,3 +16,15 @@ comments: true
 </div>
 
 :::supervision.utils.image.crop_image
+
+<div class="md-typeset">
+  <h2>resize_image</h2>
+</div>
+
+:::supervision.utils.image.resize_image
+
+<div class="md-typeset">
+  <h2>place_image</h2>
+</div>
+
+:::supervision.utils.image.place_image

docs/utils/notebook.md

@@ -2,7 +2,7 @@
 comments: true
 ---
 
-# Notebooks
+# Notebooks Utils
 
 <div class="md-typeset">
   <h2>plot_image</h2>

docs/utils/video.md

@@ -2,7 +2,7 @@
 comments: true
 ---
 
-# Video
+# Video Utils
 
 <div class="md-typeset">
   <h2>VideoInfo</h2>

examples/README.md

@@ -8,6 +8,7 @@ interfaces with diverse applications.
 - [count people in zone](./count_people_in_zone) by [@SkalskiP](https://github.com/SkalskiP)
 - [traffic analysis](./traffic_analysis) by [@SkalskiP](https://github.com/SkalskiP)
 - [speed estimation](./speed_estimation) by [@SkalskiP](https://github.com/SkalskiP)
+- [time in zone](./time_in_zone) by [@SkalskiP](https://github.com/SkalskiP)
 - [heatmap and track](./heatmap_and_track/) by [@HinePo](https://github.com/HinePo)
 
 ## Contributing

examples/count_people_in_zone/requirements.txt

@@ -1,5 +1,5 @@
 gdown
 inference
-supervision
+supervision==0.19.0
 tqdm
 ultralytics

examples/heatmap_and_track/requirements.txt

@@ -1,2 +1,2 @@
-supervision[assets]
+supervision[assets]==0.19.0
 ultralytics

examples/speed_estimation/requirements.txt

@@ -1,4 +1,4 @@
-supervision==0.18.0rc1
+supervision==0.19.0
 tqdm==4.66.1
 requests
 ultralytics==8.0.237

examples/time_in_zone/.gitignore

@@ -0,0 +1,9 @@
+data/
+venv*/
+*.pt
+*.pth
+*.mp4
+*.mov
+*.png
+*.jpg
+*.jpeg

examples/time_in_zone/README.md

@@ -0,0 +1,262 @@
+# time in zone
+
+## 👋 hello
+
+Practical demonstration on leveraging computer vision for analyzing wait times and
+monitoring the duration that objects or individuals spend in predefined areas of video
+frames. This example project, perfect for retail analytics or traffic management
+applications.
+
+https://github.com/roboflow/supervision/assets/26109316/d051cc8a-dd15-41d4-aa36-d38b86334c39
+
+## 💻 install
+
+- clone repository and navigate to example directory
+
+  ```bash
+  git clone https://github.com/roboflow/supervision.git
+  cd supervision/examples/time_in_zone
+  ```
+
+- setup python environment and activate it [optional]
+
+  ```bash
+  python3 -m venv venv
+  source venv/bin/activate
+  ```
+
+- install required dependencies
+
+  ```bash
+  pip install -r requirements.txt
+  ```
+
+## 🛠 scripts
+
+### `download_from_youtube`
+
+This script allows you to download a video from YouTube.
+
+- `--url`: The full URL of the YouTube video you wish to download.
+- `--output_path` (optional): Specifies the directory where the video will be saved.
+- `--file_name` (optional): Sets the name of the saved video file.
+
+```bash
+python scripts/download_from_youtube.py \
+--url "https://www.youtube.com/watch?v=-8zyEwAa50Q" \
+--output_path "data/checkout" \
+--file_name "video.mp4"
+```
+
+```bash
+python scripts/download_from_youtube.py \
+--url "https://www.youtube.com/watch?v=MNn9qKG2UFI" \
+--output_path "data/traffic" \
+--file_name "video.mp4"
+```
+
+### `stream_from_file`
+
+This script allows you to stream video files from a directory. It's an awesome way to
+mock a live video stream for local testing. Video will be streamed in a loop under
+`rtsp://localhost:8554/live0.stream` URL. This script requires docker to be installed.
+
+- `--video_directory`: Directory containing video files to stream.
+- `--number_of_streams`: Number of video files to stream.
+
+```bash
+python scripts/stream_from_file.py \
+--video_directory "data/checkout" \
+--number_of_streams 1
+```
+
+```bash
+python scripts/stream_from_file.py \
+--video_directory "data/traffic" \
+--number_of_streams 1
+```
+
+### `draw_zones`
+
+If you want to test zone time in zone analysis on your own video, you can use this
+script to design custom zones and save results as a JSON file. The script will open a
+window where you can draw polygons on the source image or video file. The polygons will
+be saved as a JSON file.
+
+- `--source_path`: Path to the source image or video file for drawing polygons.
+- `--zone_configuration_path`: Path where the polygon annotations will be saved as a JSON file.
+
+
+- `enter` - finish drawing the current polygon.
+- `escape` - cancel drawing the current polygon.
+- `q` - quit the drawing window.
+- `s` - save zone configuration to a JSON file.
+
+```bash
+python scripts/draw_zones.py \
+--source_path "data/checkout/video.mp4" \
+--zone_configuration_path "data/checkout/config.json"
+```
+
+```bash
+python scripts/draw_zones.py \
+--source_path "data/traffic/video.mp4" \
+--zone_configuration_path "data/traffic/custom_config.json"
+```
+
+https://github.com/roboflow/supervision/assets/26109316/9d514c9e-2a61-418b-ae49-6ac1ad6ae5ac
+
+## 🎬 video & stream processing
+
+### `inference_file_example`
+
+Script to run object detection on a video file using the Roboflow Inference model.
+
+  - `--zone_configuration_path`: Path to the zone configuration JSON file.
+  - `--source_video_path`: Path to the source video file.
+  - `--model_id`: Roboflow model ID.
+  - `--classes`: List of class IDs to track. If empty, all classes are tracked.
+  - `--confidence_threshold`: Confidence level for detections (`0` to `1`). Default is `0.3`.
+  - `--iou_threshold`: IOU threshold for non-max suppression. Default is `0.7`.
+
+```bash
+python inference_file_example.py \
+--zone_configuration_path "data/checkout/config.json" \
+--source_video_path "data/checkout/video.mp4" \
+--model_id "yolov8x-640" \
+--classes 0 \
+--confidence_threshold 0.3 \
+--iou_threshold 0.7
+```
+
+https://github.com/roboflow/supervision/assets/26109316/d051cc8a-dd15-41d4-aa36-d38b86334c39
+
+```bash
+python inference_file_example.py \
+--zone_configuration_path "data/traffic/config.json" \
+--source_video_path "data/traffic/video.mp4" \
+--model_id "yolov8x-640" \
+--classes 2 5 6 7 \
+--confidence_threshold 0.3 \
+--iou_threshold 0.7
+```
+
+https://github.com/roboflow/supervision/assets/26109316/5ec896d7-4b39-4426-8979-11e71666878b
+
+### `inference_stream_example`
+
+Script to run object detection on a video stream using the Roboflow Inference model.
+
+  - `--zone_configuration_path`: Path to the zone configuration JSON file.
+  - `--rtsp_url`: Complete RTSP URL for the video stream.
+  - `--model_id`: Roboflow model ID.
+  - `--classes`: List of class IDs to track. If empty, all classes are tracked.
+  - `--confidence_threshold`: Confidence level for detections (`0` to `1`). Default is `0.3`.
+  - `--iou_threshold`: IOU threshold for non-max suppression. Default is `0.7`.
+
+```bash
+python inference_file_example.py \
+--zone_configuration_path "data/checkout/config.json" \
+--rtsp_url "rtsp://localhost:8554/live0.stream" \
+--model_id "yolov8x-640" \
+--classes 0 \
+--confidence_threshold 0.3 \
+--iou_threshold 0.7
+```
+
+```bash
+python inference_file_example.py \
+--zone_configuration_path "data/traffic/config.json" \
+--rtsp_url "rtsp://localhost:8554/live0.stream" \
+--model_id "yolov8x-640" \
+--classes 2 5 6 7 \
+--confidence_threshold 0.3 \
+--iou_threshold 0.7
+```
+
+<details>
+<summary>👉 show ultralytics examples</summary>
+
+### `ultralytics_file_example`
+
+Script to run object detection on a video file using the Ultralytics YOLOv8 model.
+
+  - `--zone_configuration_path`: Path to the zone configuration JSON file.
+  - `--source_video_path`: Path to the source video file.
+  - `--weights`: Path to the model weights file. Default is `'yolov8s.pt'`.
+  - `--device`: Computation device (`'cpu'`, `'mps'` or `'cuda'`). Default is `'cpu'`.
+  - `--classes`: List of class IDs to track. If empty, all classes are tracked.
+  - `--confidence_threshold`: Confidence level for detections (`0` to `1`). Default is `0.3`.
+  - `--iou_threshold`: IOU threshold for non-max suppression. Default is `0.7`.
+
+```bash
+python inference_file_example.py \
+--zone_configuration_path "data/checkout/config.json" \
+--source_video_path "data/checkout/video.mp4" \
+--weights "yolov8x.pt" \
+--device "cpu" \
+--classes 0 \
+--confidence_threshold 0.3 \
+--iou_threshold 0.7
+```
+
+```bash
+python inference_file_example.py \
+--zone_configuration_path "data/traffic/config.json" \
+--source_video_path "data/traffic/video.mp4" \
+--weights "yolov8x.pt" \
+--device "cpu" \
+--classes 2 5 6 7 \
+--confidence_threshold 0.3 \
+--iou_threshold 0.7
+```
+
+### `ultralytics_stream_example`
+
+Script to run object detection on a video stream using the Ultralytics YOLOv8 model.
+
+  - `--zone_configuration_path`: Path to the zone configuration JSON file.
+  - `--rtsp_url`: Complete RTSP URL for the video stream.
+  - `--weights`: Path to the model weights file. Default is `'yolov8s.pt'`.
+  - `--device`: Computation device (`'cpu'`, `'mps'` or `'cuda'`). Default is `'cpu'`.
+  - `--classes`: List of class IDs to track. If empty, all classes are tracked.
+  - `--confidence_threshold`: Confidence level for detections (`0` to `1`). Default is `0.3`.
+  - `--iou_threshold`: IOU threshold for non-max suppression. Default is `0.7`.
+
+```bash
+python inference_file_example.py \
+--zone_configuration_path "data/checkout/config.json" \
+--rtsp_url "rtsp://localhost:8554/live0.stream" \
+--weights "yolov8x.pt" \
+--device "cpu" \
+--classes 0 \
+--confidence_threshold 0.3 \
+--iou_threshold 0.7
+```
+
+```bash
+python inference_file_example.py \
+--zone_configuration_path "data/traffic/config.json" \
+--rtsp_url "rtsp://localhost:8554/live0.stream" \
+--weights "yolov8x.pt" \
+--device "cpu" \
+--classes 2 5 6 7 \
+--confidence_threshold 0.3 \
+--iou_threshold 0.7
+```
+
+</details>
+
+## © license
+
+This demo integrates two main components, each with its own licensing:
+
+- ultralytics: The object detection model used in this demo, YOLOv8, is distributed
+  under the [AGPL-3.0 license](https://github.com/ultralytics/ultralytics/blob/main/LICENSE).
+  You can find more details about this license here.
+
+- supervision: The analytics code that powers the zone-based analysis in this demo is
+  based on the Supervision library, which is licensed under the
+  [MIT license](https://github.com/roboflow/supervision/blob/develop/LICENSE.md). This
+  makes the Supervision part of the code fully open source and freely usable in your
+  projects.

examples/time_in_zone/inference_file_example.py

@@ -0,0 +1,132 @@
+import argparse
+from typing import List
+
+import cv2
+import numpy as np
+from inference import get_model
+from utils.general import find_in_list, load_zones_config
+from utils.timers import FPSBasedTimer
+
+import supervision as sv
+
+COLORS = sv.ColorPalette.from_hex(["#E6194B", "#3CB44B", "#FFE119", "#3C76D1"])
+COLOR_ANNOTATOR = sv.ColorAnnotator(color=COLORS)
+LABEL_ANNOTATOR = sv.LabelAnnotator(
+    color=COLORS, text_color=sv.Color.from_hex("#000000")
+)
+
+
+def main(
+    source_video_path: str,
+    zone_configuration_path: str,
+    model_id: str,
+    confidence: float,
+    iou: float,
+    classes: List[int],
+) -> None:
+    model = get_model(model_id=model_id)
+    tracker = sv.ByteTrack(minimum_matching_threshold=0.5)
+    video_info = sv.VideoInfo.from_video_path(video_path=source_video_path)
+    frames_generator = sv.get_video_frames_generator(source_video_path)
+
+    frame = next(frames_generator)
+    resolution_wh = frame.shape[1], frame.shape[0]
+
+    polygons = load_zones_config(file_path=zone_configuration_path)
+    zones = [
+        sv.PolygonZone(
+            polygon=polygon,
+            frame_resolution_wh=resolution_wh,
+            triggering_anchors=(sv.Position.CENTER,),
+        )
+        for polygon in polygons
+    ]
+    timers = [FPSBasedTimer(video_info.fps) for _ in zones]
+
+    for frame in frames_generator:
+        results = model.infer(frame, confidence=confidence, iou_threshold=iou)[0]
+        detections = sv.Detections.from_inference(results)
+        detections = detections[find_in_list(detections.class_id, classes)]
+        detections = tracker.update_with_detections(detections)
+
+        annotated_frame = frame.copy()
+
+        for idx, zone in enumerate(zones):
+            annotated_frame = sv.draw_polygon(
+                scene=annotated_frame, polygon=zone.polygon, color=COLORS.by_idx(idx)
+            )
+
+            detections_in_zone = detections[zone.trigger(detections)]
+            time_in_zone = timers[idx].tick(detections_in_zone)
+            custom_color_lookup = np.full(detections_in_zone.class_id.shape, idx)
+
+            annotated_frame = COLOR_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                custom_color_lookup=custom_color_lookup,
+            )
+            labels = [
+                f"#{tracker_id} {int(time // 60):02d}:{int(time % 60):02d}"
+                for tracker_id, time in zip(detections_in_zone.tracker_id, time_in_zone)
+            ]
+            annotated_frame = LABEL_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                labels=labels,
+                custom_color_lookup=custom_color_lookup,
+            )
+
+        cv2.imshow("Processed Video", annotated_frame)
+        if cv2.waitKey(1) & 0xFF == ord("q"):
+            break
+    cv2.destroyAllWindows()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Calculating detections dwell time in zones, using video file."
+    )
+    parser.add_argument(
+        "--zone_configuration_path",
+        type=str,
+        required=True,
+        help="Path to the zone configuration JSON file.",
+    )
+    parser.add_argument(
+        "--source_video_path",
+        type=str,
+        required=True,
+        help="Path to the source video file.",
+    )
+    parser.add_argument(
+        "--model_id", type=str, default="yolov8s-640", help="Roboflow model ID."
+    )
+    parser.add_argument(
+        "--confidence_threshold",
+        type=float,
+        default=0.3,
+        help="Confidence level for detections (0 to 1). Default is 0.3.",
+    )
+    parser.add_argument(
+        "--iou_threshold",
+        default=0.7,
+        type=float,
+        help="IOU threshold for non-max suppression. Default is 0.7.",
+    )
+    parser.add_argument(
+        "--classes",
+        nargs="*",
+        type=int,
+        default=[],
+        help="List of class IDs to track. If empty, all classes are tracked.",
+    )
+    args = parser.parse_args()
+
+    main(
+        source_video_path=args.source_video_path,
+        zone_configuration_path=args.zone_configuration_path,
+        model_id=args.model_id,
+        confidence=args.confidence_threshold,
+        iou=args.iou_threshold,
+        classes=args.classes,
+    )

examples/time_in_zone/inference_naive_stream_example.py

@@ -0,0 +1,142 @@
+import argparse
+from typing import List
+
+import cv2
+import numpy as np
+from inference import get_model
+from utils.general import find_in_list, get_stream_frames_generator, load_zones_config
+from utils.timers import ClockBasedTimer
+
+import supervision as sv
+
+COLORS = sv.ColorPalette.from_hex(["#E6194B", "#3CB44B", "#FFE119", "#3C76D1"])
+COLOR_ANNOTATOR = sv.ColorAnnotator(color=COLORS)
+LABEL_ANNOTATOR = sv.LabelAnnotator(
+    color=COLORS, text_color=sv.Color.from_hex("#000000")
+)
+
+
+def main(
+    rtsp_url: str,
+    zone_configuration_path: str,
+    model_id: str,
+    confidence: float,
+    iou: float,
+    classes: List[int],
+) -> None:
+    model = get_model(model_id=model_id)
+    tracker = sv.ByteTrack(minimum_matching_threshold=0.5)
+    frames_generator = get_stream_frames_generator(rtsp_url=rtsp_url)
+    fps_monitor = sv.FPSMonitor()
+
+    frame = next(frames_generator)
+    resolution_wh = frame.shape[1], frame.shape[0]
+
+    polygons = load_zones_config(file_path=zone_configuration_path)
+    zones = [
+        sv.PolygonZone(
+            polygon=polygon,
+            frame_resolution_wh=resolution_wh,
+            triggering_anchors=(sv.Position.CENTER,),
+        )
+        for polygon in polygons
+    ]
+    timers = [ClockBasedTimer() for _ in zones]
+
+    for frame in frames_generator:
+        fps_monitor.tick()
+        fps = fps_monitor.fps
+
+        results = model.infer(frame, confidence=confidence, iou_threshold=iou)[0]
+        detections = sv.Detections.from_inference(results)
+        detections = detections[find_in_list(detections.class_id, classes)]
+        detections = tracker.update_with_detections(detections)
+
+        annotated_frame = frame.copy()
+        annotated_frame = sv.draw_text(
+            scene=annotated_frame,
+            text=f"{fps:.1f}",
+            text_anchor=sv.Point(40, 30),
+            background_color=sv.Color.from_hex("#A351FB"),
+            text_color=sv.Color.from_hex("#000000"),
+        )
+
+        for idx, zone in enumerate(zones):
+            annotated_frame = sv.draw_polygon(
+                scene=annotated_frame, polygon=zone.polygon, color=COLORS.by_idx(idx)
+            )
+
+            detections_in_zone = detections[zone.trigger(detections)]
+            time_in_zone = timers[idx].tick(detections_in_zone)
+            custom_color_lookup = np.full(detections_in_zone.class_id.shape, idx)
+
+            annotated_frame = COLOR_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                custom_color_lookup=custom_color_lookup,
+            )
+            labels = [
+                f"#{tracker_id} {int(time // 60):02d}:{int(time % 60):02d}"
+                for tracker_id, time in zip(detections_in_zone.tracker_id, time_in_zone)
+            ]
+            annotated_frame = LABEL_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                labels=labels,
+                custom_color_lookup=custom_color_lookup,
+            )
+
+        cv2.imshow("Processed Video", annotated_frame)
+        if cv2.waitKey(1) & 0xFF == ord("q"):
+            break
+    cv2.destroyAllWindows()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Calculating detections dwell time in zones, using RTSP stream."
+    )
+    parser.add_argument(
+        "--zone_configuration_path",
+        type=str,
+        required=True,
+        help="Path to the zone configuration JSON file.",
+    )
+    parser.add_argument(
+        "--rtsp_url",
+        type=str,
+        required=True,
+        help="Complete RTSP URL for the video stream.",
+    )
+    parser.add_argument(
+        "--model_id", type=str, default="yolov8s-640", help="Roboflow model ID."
+    )
+    parser.add_argument(
+        "--confidence_threshold",
+        type=float,
+        default=0.3,
+        help="Confidence level for detections (0 to 1). Default is 0.3.",
+    )
+    parser.add_argument(
+        "--iou_threshold",
+        default=0.7,
+        type=float,
+        help="IOU threshold for non-max suppression. Default is 0.7.",
+    )
+    parser.add_argument(
+        "--classes",
+        nargs="*",
+        type=int,
+        default=[],
+        help="List of class IDs to track. If empty, all classes are tracked.",
+    )
+    args = parser.parse_args()
+
+    main(
+        rtsp_url=args.rtsp_url,
+        zone_configuration_path=args.zone_configuration_path,
+        model_id=args.model_id,
+        confidence=args.confidence_threshold,
+        iou=args.iou_threshold,
+        classes=args.classes,
+    )

examples/time_in_zone/inference_stream_example.py

@@ -0,0 +1,158 @@
+import argparse
+from typing import List
+
+import cv2
+import numpy as np
+from inference import InferencePipeline
+from inference.core.interfaces.camera.entities import VideoFrame
+from utils.general import find_in_list, load_zones_config
+from utils.timers import ClockBasedTimer
+
+import supervision as sv
+
+COLORS = sv.ColorPalette.from_hex(["#E6194B", "#3CB44B", "#FFE119", "#3C76D1"])
+COLOR_ANNOTATOR = sv.ColorAnnotator(color=COLORS)
+LABEL_ANNOTATOR = sv.LabelAnnotator(
+    color=COLORS, text_color=sv.Color.from_hex("#000000")
+)
+
+
+class CustomSink:
+    def __init__(self, zone_configuration_path: str, classes: List[int]):
+        self.classes = classes
+        self.tracker = sv.ByteTrack(minimum_matching_threshold=0.5)
+        self.fps_monitor = sv.FPSMonitor()
+        self.polygons = load_zones_config(file_path=zone_configuration_path)
+        self.timers = [ClockBasedTimer() for _ in self.polygons]
+        self.zones = None
+
+    def on_prediction(self, result: dict, frame: VideoFrame) -> None:
+        if self.zones is None:
+            resolution_wh = frame.image.shape[1], frame.image.shape[0]
+            self.zones = [
+                sv.PolygonZone(
+                    polygon=polygon,
+                    frame_resolution_wh=resolution_wh,
+                    triggering_anchors=(sv.Position.CENTER,),
+                )
+                for polygon in self.polygons
+            ]
+
+        self.fps_monitor.tick()
+        fps = self.fps_monitor.fps
+
+        detections = sv.Detections.from_inference(result)
+        detections = detections[find_in_list(detections.class_id, self.classes)]
+        detections = self.tracker.update_with_detections(detections)
+
+        annotated_frame = frame.image.copy()
+        annotated_frame = sv.draw_text(
+            scene=annotated_frame,
+            text=f"{fps:.1f}",
+            text_anchor=sv.Point(40, 30),
+            background_color=sv.Color.from_hex("#A351FB"),
+            text_color=sv.Color.from_hex("#000000"),
+        )
+
+        for idx, zone in enumerate(self.zones):
+            annotated_frame = sv.draw_polygon(
+                scene=annotated_frame, polygon=zone.polygon, color=COLORS.by_idx(idx)
+            )
+
+            detections_in_zone = detections[zone.trigger(detections)]
+            time_in_zone = self.timers[idx].tick(detections_in_zone)
+            custom_color_lookup = np.full(detections_in_zone.class_id.shape, idx)
+
+            annotated_frame = COLOR_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                custom_color_lookup=custom_color_lookup,
+            )
+            labels = [
+                f"#{tracker_id} {int(time // 60):02d}:{int(time % 60):02d}"
+                for tracker_id, time in zip(detections_in_zone.tracker_id, time_in_zone)
+            ]
+            annotated_frame = LABEL_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                labels=labels,
+                custom_color_lookup=custom_color_lookup,
+            )
+        cv2.imshow("Processed Video", annotated_frame)
+        cv2.waitKey(1)
+
+
+def main(
+    rtsp_url: str,
+    zone_configuration_path: str,
+    model_id: str,
+    confidence: float,
+    iou: float,
+    classes: List[int],
+) -> None:
+    sink = CustomSink(zone_configuration_path=zone_configuration_path, classes=classes)
+
+    pipeline = InferencePipeline.init(
+        model_id=model_id,
+        video_reference=rtsp_url,
+        on_prediction=sink.on_prediction,
+        confidence=confidence,
+        iou_threshold=iou,
+    )
+
+    pipeline.start()
+
+    try:
+        pipeline.join()
+    except KeyboardInterrupt:
+        pipeline.terminate()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Calculating detections dwell time in zones, using RTSP stream."
+    )
+    parser.add_argument(
+        "--zone_configuration_path",
+        type=str,
+        required=True,
+        help="Path to the zone configuration JSON file.",
+    )
+    parser.add_argument(
+        "--rtsp_url",
+        type=str,
+        required=True,
+        help="Complete RTSP URL for the video stream.",
+    )
+    parser.add_argument(
+        "--model_id", type=str, default="yolov8s-640", help="Roboflow model ID."
+    )
+    parser.add_argument(
+        "--confidence_threshold",
+        type=float,
+        default=0.3,
+        help="Confidence level for detections (0 to 1). Default is 0.3.",
+    )
+    parser.add_argument(
+        "--iou_threshold",
+        default=0.7,
+        type=float,
+        help="IOU threshold for non-max suppression. Default is 0.7.",
+    )
+    parser.add_argument(
+        "--classes",
+        nargs="*",
+        type=int,
+        default=[],
+        help="List of class IDs to track. If empty, all classes are tracked.",
+    )
+    args = parser.parse_args()
+
+    main(
+        rtsp_url=args.rtsp_url,
+        zone_configuration_path=args.zone_configuration_path,
+        model_id=args.model_id,
+        confidence=args.confidence_threshold,
+        iou=args.iou_threshold,
+        classes=args.classes,
+    )

examples/time_in_zone/requirements.txt

@@ -0,0 +1,5 @@
+opencv-python
+supervision
+ultralytics
+inference
+pytube

examples/time_in_zone/scripts/download_from_youtube.py

@@ -0,0 +1,46 @@
+import argparse
+import os
+from typing import Optional
+
+from pytube import YouTube
+
+
+def main(url: str, output_path: Optional[str], file_name: Optional[str]) -> None:
+    yt = YouTube(url)
+    stream = yt.streams.get_highest_resolution()
+
+    if not os.path.exists(output_path):
+        os.makedirs(output_path)
+
+    stream.download(output_path=output_path, filename=file_name)
+    final_name = file_name if file_name else yt.title
+    final_path = output_path if output_path else "current directory"
+    print(f"Download completed! Video saved as '{final_name}' in '{final_path}'.")
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Download a specific YouTube video by providing its URL."
+    )
+    parser.add_argument(
+        "--url",
+        type=str,
+        required=True,
+        help="The full URL of the YouTube video you wish to download.",
+    )
+    parser.add_argument(
+        "--output_path",
+        type=str,
+        default="data/source",
+        required=False,
+        help="Optional. Specifies the directory where the video will be saved.",
+    )
+    parser.add_argument(
+        "--file_name",
+        type=str,
+        default="video.mp4",
+        required=False,
+        help="Optional. Sets the name of the saved video file.",
+    )
+    args = parser.parse_args()
+    main(url=args.url, output_path=args.output_path, file_name=args.file_name)

examples/time_in_zone/scripts/draw_zones.py

@@ -0,0 +1,176 @@
+import argparse
+import json
+import os
+from typing import Any, Optional, Tuple
+
+import cv2
+import numpy as np
+
+import supervision as sv
+
+KEY_ENTER = 13
+KEY_NEWLINE = 10
+KEY_ESCAPE = 27
+KEY_QUIT = ord("q")
+KEY_SAVE = ord("s")
+
+THICKNESS = 2
+COLORS = sv.ColorPalette.DEFAULT
+WINDOW_NAME = "Draw Zones"
+POLYGONS = [[]]
+
+current_mouse_position: Optional[Tuple[int, int]] = None
+
+
+def resolve_source(source_path: str) -> Optional[np.ndarray]:
+    if not os.path.exists(source_path):
+        return None
+
+    image = cv2.imread(source_path)
+    if image is not None:
+        return image
+
+    frame_generator = sv.get_video_frames_generator(source_path=source_path)
+    frame = next(frame_generator)
+    return frame
+
+
+def mouse_event(event: int, x: int, y: int, flags: int, param: Any) -> None:
+    global current_mouse_position
+    if event == cv2.EVENT_MOUSEMOVE:
+        current_mouse_position = (x, y)
+    elif event == cv2.EVENT_LBUTTONDOWN:
+        POLYGONS[-1].append((x, y))
+
+
+def redraw(image: np.ndarray, original_image: np.ndarray) -> None:
+    global POLYGONS, current_mouse_position
+    image[:] = original_image.copy()
+    for idx, polygon in enumerate(POLYGONS):
+        color = (
+            COLORS.by_idx(idx).as_bgr()
+            if idx < len(POLYGONS) - 1
+            else sv.Color.WHITE.as_bgr()
+        )
+
+        if len(polygon) > 1:
+            for i in range(1, len(polygon)):
+                cv2.line(
+                    img=image,
+                    pt1=polygon[i - 1],
+                    pt2=polygon[i],
+                    color=color,
+                    thickness=THICKNESS,
+                )
+            if idx < len(POLYGONS) - 1:
+                cv2.line(
+                    img=image,
+                    pt1=polygon[-1],
+                    pt2=polygon[0],
+                    color=color,
+                    thickness=THICKNESS,
+                )
+        if idx == len(POLYGONS) - 1 and current_mouse_position is not None and polygon:
+            cv2.line(
+                img=image,
+                pt1=polygon[-1],
+                pt2=current_mouse_position,
+                color=color,
+                thickness=THICKNESS,
+            )
+    cv2.imshow(WINDOW_NAME, image)
+
+
+def close_and_finalize_polygon(image: np.ndarray, original_image: np.ndarray) -> None:
+    if len(POLYGONS[-1]) > 2:
+        cv2.line(
+            img=image,
+            pt1=POLYGONS[-1][-1],
+            pt2=POLYGONS[-1][0],
+            color=COLORS.by_idx(0).as_bgr(),
+            thickness=THICKNESS,
+        )
+    POLYGONS.append([])
+    image[:] = original_image.copy()
+    redraw_polygons(image)
+    cv2.imshow(WINDOW_NAME, image)
+
+
+def redraw_polygons(image: np.ndarray) -> None:
+    for idx, polygon in enumerate(POLYGONS[:-1]):
+        if len(polygon) > 1:
+            color = COLORS.by_idx(idx).as_bgr()
+            for i in range(len(polygon) - 1):
+                cv2.line(
+                    img=image,
+                    pt1=polygon[i],
+                    pt2=polygon[i + 1],
+                    color=color,
+                    thickness=THICKNESS,
+                )
+            cv2.line(
+                img=image,
+                pt1=polygon[-1],
+                pt2=polygon[0],
+                color=color,
+                thickness=THICKNESS,
+            )
+
+
+def save_polygons_to_json(polygons, target_path):
+    data_to_save = polygons if polygons[-1] else polygons[:-1]
+    with open(target_path, "w") as f:
+        json.dump(data_to_save, f)
+
+
+def main(source_path: str, zone_configuration_path: str) -> None:
+    global current_mouse_position
+    original_image = resolve_source(source_path=source_path)
+    if original_image is None:
+        print("Failed to load source image.")
+        return
+
+    image = original_image.copy()
+    cv2.imshow(WINDOW_NAME, image)
+    cv2.setMouseCallback(WINDOW_NAME, mouse_event, image)
+
+    while True:
+        key = cv2.waitKey(1) & 0xFF
+        if key == KEY_ENTER or key == KEY_NEWLINE:
+            close_and_finalize_polygon(image, original_image)
+        elif key == KEY_ESCAPE:
+            POLYGONS[-1] = []
+            current_mouse_position = None
+        elif key == KEY_SAVE:
+            save_polygons_to_json(POLYGONS, zone_configuration_path)
+            print(f"Polygons saved to {zone_configuration_path}")
+            break
+        redraw(image, original_image)
+        if key == KEY_QUIT:
+            break
+
+    cv2.destroyAllWindows()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Interactively draw polygons on images or video frames and save "
+        "the annotations."
+    )
+    parser.add_argument(
+        "--source_path",
+        type=str,
+        required=True,
+        help="Path to the source image or video file for drawing polygons.",
+    )
+    parser.add_argument(
+        "--zone_configuration_path",
+        type=str,
+        required=True,
+        help="Path where the polygon annotations will be saved as a JSON file.",
+    )
+    arguments = parser.parse_args()
+    main(
+        source_path=arguments.source_path,
+        zone_configuration_path=arguments.zone_configuration_path,
+    )

examples/time_in_zone/scripts/stream_from_file.py

@@ -0,0 +1,104 @@
+import argparse
+import os
+import subprocess
+import tempfile
+from glob import glob
+from threading import Thread
+
+import yaml
+
+SERVER_CONFIG = {"protocols": ["tcp"], "paths": {"all": {"source": "publisher"}}}
+BASE_STREAM_URL = "rtsp://localhost:8554/live"
+
+
+def main(video_directory: str, number_of_streams: int) -> None:
+    video_files = find_video_files_in_directory(video_directory, number_of_streams)
+    try:
+        with tempfile.TemporaryDirectory() as temporary_directory:
+            config_file_path = create_server_config_file(temporary_directory)
+            run_rtsp_server(config_path=config_file_path)
+            stream_videos(video_files)
+    finally:
+        stop_rtsp_server()
+
+
+def find_video_files_in_directory(directory: str, limit: int) -> list:
+    video_formats = ["*.mp4", "*.webm"]
+    video_paths = []
+    for video_format in video_formats:
+        video_paths.extend(glob(os.path.join(directory, video_format)))
+    return video_paths[:limit]
+
+
+def create_server_config_file(directory: str) -> str:
+    config_path = os.path.join(directory, "rtsp-simple-server.yml")
+    with open(config_path, "w") as config_file:
+        yaml.dump(SERVER_CONFIG, config_file)
+    return config_path
+
+
+def run_rtsp_server(config_path: str) -> None:
+    command = (
+        "docker run --rm --name rtsp_server -d -v "
+        f"{config_path}:/rtsp-simple-server.yml -p 8554:8554 "
+        "aler9/rtsp-simple-server:v1.3.0"
+    )
+    if run_command(command.split()) != 0:
+        raise RuntimeError("Could not start the RTSP server!")
+
+
+def stop_rtsp_server() -> None:
+    run_command("docker kill rtsp_server".split())
+
+
+def stream_videos(video_files: list) -> None:
+    threads = []
+    for index, video_file in enumerate(video_files):
+        stream_url = f"{BASE_STREAM_URL}{index}.stream"
+        print(f"Streaming {video_file} under {stream_url}")
+        thread = stream_video_to_url(video_file, stream_url)
+        threads.append(thread)
+    for thread in threads:
+        thread.join()
+
+
+def stream_video_to_url(video_path: str, stream_url: str) -> Thread:
+    command = (
+        f"ffmpeg -re -stream_loop -1 -i {video_path} "
+        f"-f rtsp -rtsp_transport tcp {stream_url}"
+    )
+    return run_command_in_thread(command.split())
+
+
+def run_command_in_thread(command: list) -> Thread:
+    thread = Thread(target=run_command, args=(command,))
+    thread.start()
+    return thread
+
+
+def run_command(command: list) -> int:
+    process = subprocess.run(command)
+    return process.returncode
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Script to stream videos using RTSP protocol."
+    )
+    parser.add_argument(
+        "--video_directory",
+        type=str,
+        required=True,
+        help="Directory containing video files to stream.",
+    )
+    parser.add_argument(
+        "--number_of_streams",
+        type=int,
+        default=6,
+        help="Number of video files to stream.",
+    )
+    arguments = parser.parse_args()
+    main(
+        video_directory=arguments.video_directory,
+        number_of_streams=arguments.number_of_streams,
+    )

examples/time_in_zone/ultralytics_file_example.py

@@ -0,0 +1,144 @@
+import argparse
+from typing import List
+
+import cv2
+import numpy as np
+from ultralytics import YOLO
+from utils.general import find_in_list, load_zones_config
+from utils.timers import FPSBasedTimer
+
+import supervision as sv
+
+COLORS = sv.ColorPalette.from_hex(["#E6194B", "#3CB44B", "#FFE119", "#3C76D1"])
+COLOR_ANNOTATOR = sv.ColorAnnotator(color=COLORS)
+LABEL_ANNOTATOR = sv.LabelAnnotator(
+    color=COLORS, text_color=sv.Color.from_hex("#000000")
+)
+
+
+def main(
+    source_video_path: str,
+    zone_configuration_path: str,
+    weights: str,
+    device: str,
+    confidence: float,
+    iou: float,
+    classes: List[int],
+) -> None:
+    model = YOLO(weights)
+    tracker = sv.ByteTrack(minimum_matching_threshold=0.5)
+    video_info = sv.VideoInfo.from_video_path(video_path=source_video_path)
+    frames_generator = sv.get_video_frames_generator(source_video_path)
+
+    frame = next(frames_generator)
+    resolution_wh = frame.shape[1], frame.shape[0]
+
+    polygons = load_zones_config(file_path=zone_configuration_path)
+    zones = [
+        sv.PolygonZone(
+            polygon=polygon,
+            frame_resolution_wh=resolution_wh,
+            triggering_anchors=(sv.Position.CENTER,),
+        )
+        for polygon in polygons
+    ]
+    timers = [FPSBasedTimer(video_info.fps) for _ in zones]
+
+    for frame in frames_generator:
+        results = model(frame, verbose=False, device=device, conf=confidence)[0]
+        detections = sv.Detections.from_ultralytics(results)
+        detections = detections[find_in_list(detections.class_id, classes)]
+        detections = detections.with_nms(threshold=iou)
+        detections = tracker.update_with_detections(detections)
+
+        annotated_frame = frame.copy()
+
+        for idx, zone in enumerate(zones):
+            annotated_frame = sv.draw_polygon(
+                scene=annotated_frame, polygon=zone.polygon, color=COLORS.by_idx(idx)
+            )
+
+            detections_in_zone = detections[zone.trigger(detections)]
+            time_in_zone = timers[idx].tick(detections_in_zone)
+            custom_color_lookup = np.full(detections_in_zone.class_id.shape, idx)
+
+            annotated_frame = COLOR_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                custom_color_lookup=custom_color_lookup,
+            )
+            labels = [
+                f"#{tracker_id} {int(time // 60):02d}:{int(time % 60):02d}"
+                for tracker_id, time in zip(detections_in_zone.tracker_id, time_in_zone)
+            ]
+            annotated_frame = LABEL_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                labels=labels,
+                custom_color_lookup=custom_color_lookup,
+            )
+
+        cv2.imshow("Processed Video", annotated_frame)
+        if cv2.waitKey(1) & 0xFF == ord("q"):
+            break
+    cv2.destroyAllWindows()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Calculating detections dwell time in zones, using video file."
+    )
+    parser.add_argument(
+        "--zone_configuration_path",
+        type=str,
+        required=True,
+        help="Path to the zone configuration JSON file.",
+    )
+    parser.add_argument(
+        "--source_video_path",
+        type=str,
+        required=True,
+        help="Path to the source video file.",
+    )
+    parser.add_argument(
+        "--weights",
+        type=str,
+        default="yolov8s.pt",
+        help="Path to the model weights file. Default is 'yolov8s.pt'.",
+    )
+    parser.add_argument(
+        "--device",
+        type=str,
+        default="cpu",
+        help="Computation device ('cpu', 'mps' or 'cuda'). Default is 'cpu'.",
+    )
+    parser.add_argument(
+        "--confidence_threshold",
+        type=float,
+        default=0.3,
+        help="Confidence level for detections (0 to 1). Default is 0.3.",
+    )
+    parser.add_argument(
+        "--iou_threshold",
+        default=0.7,
+        type=float,
+        help="IOU threshold for non-max suppression. Default is 0.7.",
+    )
+    parser.add_argument(
+        "--classes",
+        nargs="*",
+        type=int,
+        default=[],
+        help="List of class IDs to track. If empty, all classes are tracked.",
+    )
+    args = parser.parse_args()
+
+    main(
+        source_video_path=args.source_video_path,
+        zone_configuration_path=args.zone_configuration_path,
+        weights=args.weights,
+        device=args.device,
+        confidence=args.confidence_threshold,
+        iou=args.iou_threshold,
+        classes=args.classes,
+    )

examples/time_in_zone/ultralytics_naive_stream_example.py

@@ -0,0 +1,154 @@
+import argparse
+from typing import List
+
+import cv2
+import numpy as np
+from ultralytics import YOLO
+from utils.general import find_in_list, get_stream_frames_generator, load_zones_config
+from utils.timers import ClockBasedTimer
+
+import supervision as sv
+
+COLORS = sv.ColorPalette.from_hex(["#E6194B", "#3CB44B", "#FFE119", "#3C76D1"])
+COLOR_ANNOTATOR = sv.ColorAnnotator(color=COLORS)
+LABEL_ANNOTATOR = sv.LabelAnnotator(
+    color=COLORS, text_color=sv.Color.from_hex("#000000")
+)
+
+
+def main(
+    rtsp_url: str,
+    zone_configuration_path: str,
+    weights: str,
+    device: str,
+    confidence: float,
+    iou: float,
+    classes: List[int],
+) -> None:
+    model = YOLO(weights)
+    tracker = sv.ByteTrack(minimum_matching_threshold=0.5)
+    frames_generator = get_stream_frames_generator(rtsp_url=rtsp_url)
+    fps_monitor = sv.FPSMonitor()
+
+    frame = next(frames_generator)
+    resolution_wh = frame.shape[1], frame.shape[0]
+
+    polygons = load_zones_config(file_path=zone_configuration_path)
+    zones = [
+        sv.PolygonZone(
+            polygon=polygon,
+            frame_resolution_wh=resolution_wh,
+            triggering_anchors=(sv.Position.CENTER,),
+        )
+        for polygon in polygons
+    ]
+    timers = [ClockBasedTimer() for _ in zones]
+
+    for frame in frames_generator:
+        fps_monitor.tick()
+        fps = fps_monitor.fps
+
+        results = model(frame, verbose=False, device=device, conf=confidence)[0]
+        detections = sv.Detections.from_ultralytics(results)
+        detections = detections[find_in_list(detections.class_id, classes)]
+        detections = detections.with_nms(threshold=iou)
+        detections = tracker.update_with_detections(detections)
+
+        annotated_frame = frame.copy()
+        annotated_frame = sv.draw_text(
+            scene=annotated_frame,
+            text=f"{fps:.1f}",
+            text_anchor=sv.Point(40, 30),
+            background_color=sv.Color.from_hex("#A351FB"),
+            text_color=sv.Color.from_hex("#000000"),
+        )
+
+        for idx, zone in enumerate(zones):
+            annotated_frame = sv.draw_polygon(
+                scene=annotated_frame, polygon=zone.polygon, color=COLORS.by_idx(idx)
+            )
+
+            detections_in_zone = detections[zone.trigger(detections)]
+            time_in_zone = timers[idx].tick(detections_in_zone)
+            custom_color_lookup = np.full(detections_in_zone.class_id.shape, idx)
+
+            annotated_frame = COLOR_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                custom_color_lookup=custom_color_lookup,
+            )
+            labels = [
+                f"#{tracker_id} {int(time // 60):02d}:{int(time % 60):02d}"
+                for tracker_id, time in zip(detections_in_zone.tracker_id, time_in_zone)
+            ]
+            annotated_frame = LABEL_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                labels=labels,
+                custom_color_lookup=custom_color_lookup,
+            )
+
+        cv2.imshow("Processed Video", annotated_frame)
+        if cv2.waitKey(1) & 0xFF == ord("q"):
+            break
+    cv2.destroyAllWindows()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Calculating detections dwell time in zones, using RTSP stream."
+    )
+    parser.add_argument(
+        "--zone_configuration_path",
+        type=str,
+        required=True,
+        help="Path to the zone configuration JSON file.",
+    )
+    parser.add_argument(
+        "--rtsp_url",
+        type=str,
+        required=True,
+        help="Complete RTSP URL for the video stream.",
+    )
+    parser.add_argument(
+        "--weights",
+        type=str,
+        default="yolov8s.pt",
+        help="Path to the model weights file. Default is 'yolov8s.pt'.",
+    )
+    parser.add_argument(
+        "--device",
+        type=str,
+        default="cpu",
+        help="Computation device ('cpu', 'mps' or 'cuda'). Default is 'cpu'.",
+    )
+    parser.add_argument(
+        "--confidence_threshold",
+        type=float,
+        default=0.3,
+        help="Confidence level for detections (0 to 1). Default is 0.3.",
+    )
+    parser.add_argument(
+        "--iou_threshold",
+        default=0.7,
+        type=float,
+        help="IOU threshold for non-max suppression. Default is 0.7.",
+    )
+    parser.add_argument(
+        "--classes",
+        nargs="*",
+        type=int,
+        default=[],
+        help="List of class IDs to track. If empty, all classes are tracked.",
+    )
+    args = parser.parse_args()
+
+    main(
+        rtsp_url=args.rtsp_url,
+        zone_configuration_path=args.zone_configuration_path,
+        weights=args.weights,
+        device=args.device,
+        confidence=args.confidence_threshold,
+        iou=args.iou_threshold,
+        classes=args.classes,
+    )

examples/time_in_zone/ultralytics_stream_example.py

@@ -0,0 +1,173 @@
+import argparse
+from typing import List
+
+import cv2
+import numpy as np
+from inference import InferencePipeline
+from inference.core.interfaces.camera.entities import VideoFrame
+from ultralytics import YOLO
+from utils.general import find_in_list, load_zones_config
+from utils.timers import ClockBasedTimer
+
+import supervision as sv
+
+COLORS = sv.ColorPalette.from_hex(["#E6194B", "#3CB44B", "#FFE119", "#3C76D1"])
+COLOR_ANNOTATOR = sv.ColorAnnotator(color=COLORS)
+LABEL_ANNOTATOR = sv.LabelAnnotator(
+    color=COLORS, text_color=sv.Color.from_hex("#000000")
+)
+
+
+class CustomSink:
+    def __init__(self, zone_configuration_path: str, classes: List[int]):
+        self.classes = classes
+        self.tracker = sv.ByteTrack(minimum_matching_threshold=0.8)
+        self.fps_monitor = sv.FPSMonitor()
+        self.polygons = load_zones_config(file_path=zone_configuration_path)
+        self.timers = [ClockBasedTimer() for _ in self.polygons]
+        self.zones = None
+
+    def on_prediction(self, detections: sv.Detections, frame: VideoFrame) -> None:
+        if self.zones is None:
+            resolution_wh = frame.image.shape[1], frame.image.shape[0]
+            self.zones = [
+                sv.PolygonZone(
+                    polygon=polygon,
+                    frame_resolution_wh=resolution_wh,
+                    triggering_anchors=(sv.Position.CENTER,),
+                )
+                for polygon in self.polygons
+            ]
+
+        self.fps_monitor.tick()
+        fps = self.fps_monitor.fps
+
+        detections = detections[find_in_list(detections.class_id, self.classes)]
+        detections = self.tracker.update_with_detections(detections)
+
+        annotated_frame = frame.image.copy()
+        annotated_frame = sv.draw_text(
+            scene=annotated_frame,
+            text=f"{fps:.1f}",
+            text_anchor=sv.Point(40, 30),
+            background_color=sv.Color.from_hex("#A351FB"),
+            text_color=sv.Color.from_hex("#000000"),
+        )
+
+        for idx, zone in enumerate(self.zones):
+            annotated_frame = sv.draw_polygon(
+                scene=annotated_frame, polygon=zone.polygon, color=COLORS.by_idx(idx)
+            )
+
+            detections_in_zone = detections[zone.trigger(detections)]
+            time_in_zone = self.timers[idx].tick(detections_in_zone)
+            custom_color_lookup = np.full(detections_in_zone.class_id.shape, idx)
+
+            annotated_frame = COLOR_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                custom_color_lookup=custom_color_lookup,
+            )
+            labels = [
+                f"#{tracker_id} {int(time // 60):02d}:{int(time % 60):02d}"
+                for tracker_id, time in zip(detections_in_zone.tracker_id, time_in_zone)
+            ]
+            annotated_frame = LABEL_ANNOTATOR.annotate(
+                scene=annotated_frame,
+                detections=detections_in_zone,
+                labels=labels,
+                custom_color_lookup=custom_color_lookup,
+            )
+        cv2.imshow("Processed Video", annotated_frame)
+        cv2.waitKey(1)
+
+
+def main(
+    rtsp_url: str,
+    zone_configuration_path: str,
+    weights: str,
+    device: str,
+    confidence: float,
+    iou: float,
+    classes: List[int],
+) -> None:
+    model = YOLO(weights)
+
+    def inference_callback(frame: VideoFrame) -> sv.Detections:
+        results = model(frame.image, verbose=False, conf=confidence, device=device)[0]
+        return sv.Detections.from_ultralytics(results).with_nms(threshold=iou)
+
+    sink = CustomSink(zone_configuration_path=zone_configuration_path, classes=classes)
+
+    pipeline = InferencePipeline.init_with_custom_logic(
+        video_reference=rtsp_url,
+        on_video_frame=inference_callback,
+        on_prediction=sink.on_prediction,
+    )
+
+    pipeline.start()
+
+    try:
+        pipeline.join()
+    except KeyboardInterrupt:
+        pipeline.terminate()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Calculating detections dwell time in zones, using RTSP stream."
+    )
+    parser.add_argument(
+        "--zone_configuration_path",
+        type=str,
+        required=True,
+        help="Path to the zone configuration JSON file.",
+    )
+    parser.add_argument(
+        "--rtsp_url",
+        type=str,
+        required=True,
+        help="Complete RTSP URL for the video stream.",
+    )
+    parser.add_argument(
+        "--weights",
+        type=str,
+        default="yolov8s.pt",
+        help="Path to the model weights file. Default is 'yolov8s.pt'.",
+    )
+    parser.add_argument(
+        "--device",
+        type=str,
+        default="cpu",
+        help="Computation device ('cpu', 'mps' or 'cuda'). Default is 'cpu'.",
+    )
+    parser.add_argument(
+        "--confidence_threshold",
+        type=float,
+        default=0.3,
+        help="Confidence level for detections (0 to 1). Default is 0.3.",
+    )
+    parser.add_argument(
+        "--iou_threshold",
+        default=0.7,
+        type=float,
+        help="IOU threshold for non-max suppression. Default is 0.7.",
+    )
+    parser.add_argument(
+        "--classes",
+        nargs="*",
+        type=int,
+        default=[],
+        help="List of class IDs to track. If empty, all classes are tracked.",
+    )
+    args = parser.parse_args()
+
+    main(
+        rtsp_url=args.rtsp_url,
+        zone_configuration_path=args.zone_configuration_path,
+        weights=args.weights,
+        device=args.device,
+        confidence=args.confidence_threshold,
+        iou=args.iou_threshold,
+        classes=args.classes,
+    )

examples/time_in_zone/utils/general.py

@@ -0,0 +1,66 @@
+import json
+from typing import Generator, List
+
+import cv2
+import numpy as np
+
+
+def load_zones_config(file_path: str) -> List[np.ndarray]:
+    """
+    Load polygon zone configurations from a JSON file.
+
+    This function reads a JSON file which contains polygon coordinates, and
+    converts them into a list of NumPy arrays. Each polygon is represented as
+    a NumPy array of coordinates.
+
+    Args:
+        file_path (str): The path to the JSON configuration file.
+
+    Returns:
+        List[np.ndarray]: A list of polygons, each represented as a NumPy array.
+    """
+    with open(file_path, "r") as file:
+        data = json.load(file)
+        return [np.array(polygon, np.int32) for polygon in data]
+
+
+def find_in_list(array: np.ndarray, search_list: List[int]) -> np.ndarray:
+    """Determines if elements of a numpy array are present in a list.
+
+    Args:
+        array (np.ndarray): The numpy array of integers to check.
+        search_list (List[int]): The list of integers to search within.
+
+    Returns:
+        np.ndarray: A numpy array of booleans, where each boolean indicates whether
+        the corresponding element in `array` is found in `search_list`.
+    """
+    if not search_list:
+        return np.ones(array.shape, dtype=bool)
+    else:
+        return np.isin(array, search_list)
+
+
+def get_stream_frames_generator(rtsp_url: str) -> Generator[np.ndarray, None, None]:
+    """
+    Generator function to yield frames from an RTSP stream.
+
+    Args:
+        rtsp_url (str): URL of the RTSP video stream.
+
+    Yields:
+        np.ndarray: The next frame from the video stream.
+    """
+    cap = cv2.VideoCapture(rtsp_url)
+    if not cap.isOpened():
+        raise Exception("Error: Could not open video stream.")
+
+    try:
+        while True:
+            ret, frame = cap.read()
+            if not ret:
+                print("End of stream or error reading frame.")
+                break
+            yield frame
+    finally:
+        cap.release()

examples/time_in_zone/utils/timers.py

@@ -0,0 +1,88 @@
+from datetime import datetime
+from typing import Dict
+
+import numpy as np
+
+import supervision as sv
+
+
+class FPSBasedTimer:
+    """
+    A timer that calculates the duration each object has been detected based on frames
+    per second (FPS).
+
+    Attributes:
+        fps (int): The frame rate of the video stream, used to calculate time durations.
+        frame_id (int): The current frame number in the sequence.
+        tracker_id2frame_id (Dict[int, int]): Maps each tracker's ID to the frame number
+            at which it was first detected.
+    """
+
+    def __init__(self, fps: int = 30) -> None:
+        """Initializes the FPSBasedTimer with the specified frames per second rate.
+
+        Args:
+            fps (int, optional): The frame rate of the video stream. Defaults to 30.
+        """
+        self.fps = fps
+        self.frame_id = 0
+        self.tracker_id2frame_id: Dict[int, int] = {}
+
+    def tick(self, detections: sv.Detections) -> np.ndarray:
+        """Processes the current frame, updating time durations for each tracker.
+
+        Args:
+            detections: The detections for the current frame, including tracker IDs.
+
+        Returns:
+            np.ndarray: Time durations (in seconds) for each detected tracker, since
+                their first detection.
+        """
+        self.frame_id += 1
+        times = []
+
+        for tracker_id in detections.tracker_id:
+            self.tracker_id2frame_id.setdefault(tracker_id, self.frame_id)
+
+            start_frame_id = self.tracker_id2frame_id[tracker_id]
+            time_duration = (self.frame_id - start_frame_id) / self.fps
+            times.append(time_duration)
+
+        return np.array(times)
+
+
+class ClockBasedTimer:
+    """
+    A timer that calculates the duration each object has been detected based on the
+    system clock.
+
+    Attributes:
+        tracker_id2start_time (Dict[int, datetime]): Maps each tracker's ID to the
+            datetime when it was first detected.
+    """
+
+    def __init__(self) -> None:
+        """Initializes the ClockBasedTimer."""
+        self.tracker_id2start_time: Dict[int, datetime] = {}
+
+    def tick(self, detections: sv.Detections) -> np.ndarray:
+        """Processes the current frame, updating time durations for each tracker.
+
+        Args:
+            detections: The detections for the current frame, including tracker IDs.
+
+        Returns:
+            np.ndarray: Time durations (in seconds) for each detected tracker, since
+                their first detection.
+        """
+        current_time = datetime.now()
+        times = []
+
+        for tracker_id in detections.tracker_id:
+            self.tracker_id2start_time.setdefault(tracker_id, current_time)
+
+            start_time = self.tracker_id2start_time[tracker_id]
+            time_duration = (current_time - start_time).total_seconds()
+            times.append(time_duration)
+
+        return np.array(times)

examples/tracking/requirements.txt

@@ -1,4 +1,4 @@
 inference
-supervision
+supervision==0.19.0
 tqdm
 ultralytics

examples/traffic_analysis/inference_example.py

@@ -1,6 +1,6 @@
 import argparse
 import os
-from typing import Dict, List, Set, Tuple
+from typing import Dict, Iterable, List, Set, Tuple
 
 import cv2
 import numpy as np
@@ -9,7 +9,8 @@ from tqdm import tqdm
 
 import supervision as sv
 
-COLORS = sv.ColorPalette.default()
+COLORS = sv.ColorPalette.from_hex(["#E6194B", "#3CB44B", "#FFE119", "#3C76D1"])
+
 
 ZONE_IN_POLYGONS = [
     np.array([[592, 282], [900, 282], [900, 82], [592, 82]]),
@@ -60,13 +61,13 @@ class DetectionsManager:
 def initiate_polygon_zones(
     polygons: List[np.ndarray],
     frame_resolution_wh: Tuple[int, int],
-    triggering_position: sv.Position = sv.Position.CENTER,
+    triggering_anchors: Iterable[sv.Position] = [sv.Position.CENTER],
 ) -> List[sv.PolygonZone]:
     return [
         sv.PolygonZone(
             polygon=polygon,
             frame_resolution_wh=frame_resolution_wh,
-            triggering_position=triggering_position,
+            triggering_anchors=triggering_anchors,
         )
         for polygon in polygons
     ]
@@ -92,13 +93,16 @@ class VideoProcessor:
 
         self.video_info = sv.VideoInfo.from_video_path(source_video_path)
         self.zones_in = initiate_polygon_zones(
-            ZONE_IN_POLYGONS, self.video_info.resolution_wh, sv.Position.CENTER
+            ZONE_IN_POLYGONS, self.video_info.resolution_wh, [sv.Position.CENTER]
         )
         self.zones_out = initiate_polygon_zones(
-            ZONE_OUT_POLYGONS, self.video_info.resolution_wh, sv.Position.CENTER
+            ZONE_OUT_POLYGONS, self.video_info.resolution_wh, [sv.Position.CENTER]
         )
 
-        self.box_annotator = sv.BoxAnnotator(color=COLORS)
+        self.bounding_box_annotator = sv.BoundingBoxAnnotator(color=COLORS)
+        self.label_annotator = sv.LabelAnnotator(
+            color=COLORS, text_color=sv.Color.BLACK
+        )
         self.trace_annotator = sv.TraceAnnotator(
             color=COLORS, position=sv.Position.CENTER, trace_length=100, thickness=2
         )
@@ -136,7 +140,10 @@ class VideoProcessor:
 
         labels = [f"#{tracker_id}" for tracker_id in detections.tracker_id]
         annotated_frame = self.trace_annotator.annotate(annotated_frame, detections)
-        annotated_frame = self.box_annotator.annotate(
+        annotated_frame = self.bounding_box_annotator.annotate(
+            annotated_frame, detections
+        )
+        annotated_frame = self.label_annotator.annotate(
             annotated_frame, detections, labels
         )
 
@@ -167,7 +174,7 @@ class VideoProcessor:
         detections_in_zones = []
         detections_out_zones = []
 
-        for i, (zone_in, zone_out) in enumerate(zip(self.zones_in, self.zones_out)):
+        for zone_in, zone_out in zip(self.zones_in, self.zones_out):
             detections_in_zone = detections[zone_in.trigger(detections=detections)]
             detections_in_zones.append(detections_in_zone)
             detections_out_zone = detections[zone_out.trigger(detections=detections)]

examples/traffic_analysis/requirements.txt

@@ -1,5 +1,5 @@
 gdown
 inference
-supervision
+supervision>=0.19.0
 tqdm
 ultralytics

examples/traffic_analysis/ultralytics_example.py

@@ -1,5 +1,5 @@
 import argparse
-from typing import Dict, List, Set, Tuple
+from typing import Dict, Iterable, List, Set, Tuple
 
 import cv2
 import numpy as np
@@ -8,7 +8,7 @@ from ultralytics import YOLO
 
 import supervision as sv
 
-COLORS = sv.ColorPalette.default()
+COLORS = sv.ColorPalette.from_hex(["#E6194B", "#3CB44B", "#FFE119", "#3C76D1"])
 
 ZONE_IN_POLYGONS = [
     np.array([[592, 282], [900, 282], [900, 82], [592, 82]]),
@@ -59,13 +59,13 @@ class DetectionsManager:
 def initiate_polygon_zones(
     polygons: List[np.ndarray],
     frame_resolution_wh: Tuple[int, int],
-    triggering_position: sv.Position = sv.Position.CENTER,
+    triggering_anchors: Iterable[sv.Position] = [sv.Position.CENTER],
 ) -> List[sv.PolygonZone]:
     return [
         sv.PolygonZone(
             polygon=polygon,
             frame_resolution_wh=frame_resolution_wh,
-            triggering_position=triggering_position,
+            triggering_anchors=triggering_anchors,
         )
         for polygon in polygons
     ]
@@ -90,13 +90,16 @@ class VideoProcessor:
 
         self.video_info = sv.VideoInfo.from_video_path(source_video_path)
         self.zones_in = initiate_polygon_zones(
-            ZONE_IN_POLYGONS, self.video_info.resolution_wh, sv.Position.CENTER
+            ZONE_IN_POLYGONS, self.video_info.resolution_wh, [sv.Position.CENTER]
         )
         self.zones_out = initiate_polygon_zones(
-            ZONE_OUT_POLYGONS, self.video_info.resolution_wh, sv.Position.CENTER
+            ZONE_OUT_POLYGONS, self.video_info.resolution_wh, [sv.Position.CENTER]
         )
 
-        self.box_annotator = sv.BoxAnnotator(color=COLORS)
+        self.bounding_box_annotator = sv.BoundingBoxAnnotator(color=COLORS)
+        self.label_annotator = sv.LabelAnnotator(
+            color=COLORS, text_color=sv.Color.BLACK
+        )
         self.trace_annotator = sv.TraceAnnotator(
             color=COLORS, position=sv.Position.CENTER, trace_length=100, thickness=2
         )
@@ -134,7 +137,10 @@ class VideoProcessor:
 
         labels = [f"#{tracker_id}" for tracker_id in detections.tracker_id]
         annotated_frame = self.trace_annotator.annotate(annotated_frame, detections)
-        annotated_frame = self.box_annotator.annotate(
+        annotated_frame = self.bounding_box_annotator.annotate(
+            annotated_frame, detections
+        )
+        annotated_frame = self.label_annotator.annotate(
             annotated_frame, detections, labels
         )
 
@@ -165,7 +171,7 @@ class VideoProcessor:
         detections_in_zones = []
         detections_out_zones = []
 
-        for i, (zone_in, zone_out) in enumerate(zip(self.zones_in, self.zones_out)):
+        for zone_in, zone_out in zip(self.zones_in, self.zones_out):
             detections_in_zone = detections[zone_in.trigger(detections=detections)]
             detections_in_zones.append(detections_in_zone)
             detections_out_zone = detections[zone_out.trigger(detections=detections)]

mkdocs.yml

@@ -31,14 +31,18 @@ extra:
 
 extra_css:
   - stylesheets/extra.css
+  - stylesheets/cookbooks-card.css
+
 
 nav:
   - Home: index.md
   - How to:
       - Detect and Annotate: how_to/detect_and_annotate.md
+      - Detect Small Objects: how_to/detect_small_objects.md
       - Track Objects: how_to/track_objects.md
       - Filter Detections: how_to/filter_detections.md
   - API:
+    - Annotators: annotators.md
     - Classifications:
         - Core: classification/core.md
     - Detections:
@@ -50,7 +54,6 @@ nav:
         - Inference Slicer: detection/tools/inference_slicer.md
         - Detection Smoother: detection/tools/smoother.md
         - Save Detections: detection/tools/save_detections.md
-    - Annotators: annotators.md
     - Trackers: trackers.md
     - Datasets: datasets.md
     - Metrics:
@@ -65,21 +68,33 @@ nav:
         - Image: utils/image.md
         - Notebook: utils/notebook.md
         - File: utils/file.md
-  - Assets: assets.md
-  - Development:
+    - Assets: assets.md
+  - Cookbooks: cookbooks.md
+  - Contribute:
     - Contributing: contributing.md
     - Code of Conduct: code_of_conduct.md
+    - License: license.md
   - Changelog:
         - Changelog: changelog.md
         - Deprecated: deprecated.md
-  - License: license.md
+
 
 
 theme:
   name: 'material'
+  icon:
+    edit: material/pencil
   logo: assets/supervision-lenny.png
   favicon: assets/supervision-lenny.png
   custom_dir: docs/theme
+  features:
+    - navigation.tracking
+    - content.code.copy
+    - content.action.edit
+    - content.tooltips
+    - content.code.annotate
+    - navigation.tabs
+
   palette:
     # Palette for light mode
     - scheme: default
@@ -94,15 +109,22 @@ theme:
       toggle:
         icon: material/brightness-4
         name: Switch to light mode
+
+
   font:
     text: Roboto
     code: Roboto Mono
-  features:
-    - content.code.copy
-    - content.code.annotate
+    features:
+      - content.code.copy
+      - content.code.annotate
 
 plugins:
   - search
+  - mkdocs-jupyter:
+        kernel_name: python3
+        execute: false
+        include_source: True
+        include_requirejs: true
   - mkdocstrings:
       default_handler: python
       handlers:
@@ -118,9 +140,15 @@ plugins:
             show_symbol_type_heading: true
             show_symbol_type_toc: true
             show_category_heading: true
-          import:
-          - https://docs.python-requests.org/en/master/objects.inv
           domains: [std, py]
+  - git-committers:
+      repository: roboflow/supervision
+      branch: develop
+      token: !ENV ["GITHUB_TOKEN"]
+  - git-revision-date-localized:
+      enable_creation_date: true
+
+
 
 markdown_extensions:
   - admonition
@@ -137,7 +165,11 @@ markdown_extensions:
       emoji_generator: !!python/name:material.extensions.emoji.to_svg
   - pymdownx.snippets:
       check_paths: true
+  - pymdownx.highlight:
+      anchor_linenums: true
 
 extra_javascript:
   - "https://widget.kapa.ai/kapa-widget.bundle.js"
-  - "javascript/init_kapa_widget.js"
+  - "javascripts/init_kapa_widget.js"
+  - "javascripts/cookbooks-card.js"
+  - "https://cdnjs.cloudflare.com/ajax/libs/dompurify/3.0.8/purify.min.js"

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "supervision"
-version = "0.19.0rc2"
+version = "0.20.0rc2"
 description = "A set of easy-to-use utils that will come in handy in any Computer Vision project"
 authors = ["Piotr Skalski <piotr.skalski92@gmail.com>"]
 maintainers = ["Piotr Skalski <piotr.skalski92@gmail.com>"]
@@ -39,35 +39,42 @@ scipy = [
 ]
 matplotlib = ">=3.6.0"
 pyyaml = ">=5.3"
+defusedxml = "^0.7.1"
 opencv-python = { version = ">=4.5.5.64", optional = true }
 opencv-python-headless = ">=4.5.5.64"
 requests = { version = ">=2.26.0,<=2.31.0", optional = true }
-tqdm = { version = ">=4.62.3,<=4.66.1", optional = true }
-defusedxml = "^0.7.1"
-
-
+tqdm = { version = ">=4.62.3,<=4.66.2", optional = true }
+pillow = ">=9.4"
 
 [tool.poetry.extras]
 desktop = ["opencv-python"]
 assets = ["requests","tqdm"]
 
 [tool.poetry.group.dev.dependencies]
-twine = "^4.0.2"
+twine = ">=4.0.2,<6.0.0"
 pytest = ">=7.2.2,<9.0.0"
-wheel = ">=0.40,<0.43"
-notebook = ">=6.5.3,<8.0.0"
-build = ">=0.10,<1.1"
+wheel = ">=0.40,<0.44"
+build = ">=0.10,<1.3"
 ruff = ">=0.1.0"
-isort = "^5.12.0"
 mypy = "^1.4.1"
 pre-commit = "^3.3.3"
 tox = "^4.11.4"
+notebook = ">=6.5.3,<8.0.0"
+ipywidgets = "^8.1.1"
+jupytext = "^1.16.1"
+nbconvert = "^7.14.2"
 
 
 [tool.poetry.group.docs.dependencies]
 mkdocs-material = {extras = ["imaging"], version = "^9.5.5"}
 mkdocstrings = {extras = ["python"], version = ">=0.20,<0.25"}
 mike = "^2.0.0"
+# For Documentation Development use Python 3.10 or above
+# Use Latest mkdocs-jupyter min 0.24.6 for Jupyter Notebook Theme support
+mkdocs-jupyter = "^0.24.3"
+mkdocs-git-committers-plugin-2 = "^2.2.3"
+mkdocs-git-revision-date-localized-plugin = "^1.2.4"
+
 
 
 [tool.isort]
@@ -104,13 +111,6 @@ exclude = '''
 
 [tool.ruff]
 target-version = "py38"
-# Enable pycodestyle (`E`) and Pyflakes (`F`) codes by default.
-select = ["E", "F"]
-ignore = []
-
-# Allow autofix for all enabled rules (when `--fix`) is provided.
-fixable = ["A", "B", "C", "D", "E", "F", "G", "I", "N", "Q", "S", "T", "W", "ANN", "ARG", "BLE", "COM", "DJ", "DTZ", "EM", "ERA", "EXE", "FBT", "ICN", "INP", "ISC", "NPY", "PD", "PGH", "PIE", "PL", "PT", "PTH", "PYI", "RET", "RSE", "RUF", "SIM", "SLF", "TCH", "TID", "TRY", "UP", "YTT"]
-unfixable = []
 
 # Exclude a variety of commonly ignored directories.
 exclude = [
@@ -140,33 +140,35 @@ exclude = [
     "docs",
 ]
 
-# Same as Black.
 line-length = 88
 indent-width = 4
 
+[tool.ruff.lint]
+# Enable pycodestyle (`E`) and Pyflakes (`F`) codes by default.
+select = ["E", "F"]
+ignore = []
+# Allow autofix for all enabled rules (when `--fix`) is provided.
+fixable = ["A", "B", "C", "D", "E", "F", "G", "I", "N", "Q", "S", "T", "W", "ANN", "ARG", "BLE", "COM", "DJ", "DTZ", "EM", "ERA", "EXE", "FBT", "ICN", "INP", "ISC", "NPY", "PD", "PGH", "PIE", "PL", "PT", "PTH", "PYI", "RET", "RSE", "RUF", "SIM", "SLF", "TCH", "TID", "TRY", "UP", "YTT"]
+unfixable = []
 # Allow unused variables when underscore-prefixed.
 dummy-variable-rgx = "^(_+|(_+[a-zA-Z0-9_]*[a-zA-Z0-9]+?))$"
+pylint.max-args = 20
 
-[tool.ruff.flake8-quotes]
+[tool.ruff.lint.flake8-quotes]
 inline-quotes = "double"
 multiline-quotes = "double"
 docstring-quotes = "double"
 
-[tool.ruff.pydocstyle]
+[tool.ruff.lint.pydocstyle]
 convention = "google"
 
-[tool.ruff.per-file-ignores]
+[tool.ruff.lint.per-file-ignores]
 "__init__.py" = ["E402","F401"]
-"supervision/assets/list.py" = ["E501"]
 
 [tool.ruff.lint.mccabe]
 # Flag errors (`C901`) whenever the complexity level exceeds 5.
 max-complexity = 20
 
-
-[tool.ruff.pylint]
-max-args = 20
-
 [tool.ruff.format]
 # Like Black, use double quotes for strings.
 quote-style = "double"

supervision/__init__.py

@@ -12,6 +12,7 @@ from supervision.annotators.core import (
     BoxCornerAnnotator,
     CircleAnnotator,
     ColorAnnotator,
+    CropAnnotator,
     DotAnnotator,
     EllipseAnnotator,
     HaloAnnotator,
@@ -35,27 +36,30 @@ from supervision.dataset.core import (
 )
 from supervision.detection.annotate import BoxAnnotator
 from supervision.detection.core import Detections
-from supervision.detection.line_counter import LineZone, LineZoneAnnotator
+from supervision.detection.line_zone import LineZone, LineZoneAnnotator
 from supervision.detection.tools.csv_sink import CSVSink
 from supervision.detection.tools.inference_slicer import InferenceSlicer
+from supervision.detection.tools.json_sink import JSONSink
 from supervision.detection.tools.polygon_zone import PolygonZone, PolygonZoneAnnotator
 from supervision.detection.tools.smoother import DetectionsSmoother
 from supervision.detection.utils import (
     box_iou_batch,
+    box_non_max_suppression,
     calculate_masks_centroids,
     filter_polygons_by_area,
+    mask_iou_batch,
+    mask_non_max_suppression,
     mask_to_polygons,
     mask_to_xyxy,
     move_boxes,
-    non_max_suppression,
     polygon_to_mask,
     polygon_to_xyxy,
     scale_boxes,
 )
 from supervision.draw.color import Color, ColorPalette
 from supervision.draw.utils import (
-    calculate_dynamic_line_thickness,
-    calculate_dynamic_text_scale,
+    calculate_optimal_line_thickness,
+    calculate_optimal_text_scale,
     draw_filled_rectangle,
     draw_image,
     draw_line,
@@ -68,7 +72,7 @@ from supervision.geometry.utils import get_polygon_center
 from supervision.metrics.detection import ConfusionMatrix, MeanAveragePrecision
 from supervision.tracker.byte_tracker.core import ByteTrack
 from supervision.utils.file import list_files_with_extensions
-from supervision.utils.image import ImageSink, crop_image
+from supervision.utils.image import ImageSink, crop_image, place_image, resize_image
 from supervision.utils.notebook import plot_image, plot_images_grid
 from supervision.utils.video import (
     FPSMonitor,

supervision/annotators/base.py

@@ -1,11 +1,22 @@
 from abc import ABC, abstractmethod
+from typing import TypeVar
 
 import numpy as np
+from PIL import Image
 
 from supervision.detection.core import Detections
 
+ImageType = TypeVar("ImageType", np.ndarray, Image.Image)
+"""
+An image of type `np.ndarray` or `PIL.Image.Image`.
+
+Unlike a `Union`, ensures the type remains consistent. If a function
+takes an `ImageType` argument and returns an `ImageType`, when you
+pass an `np.ndarray`, you will get an `np.ndarray` back.
+"""
+
 
 class BaseAnnotator(ABC):
     @abstractmethod
-    def annotate(self, scene: np.ndarray, detections: Detections) -> np.ndarray:
+    def annotate(self, scene: ImageType, detections: Detections) -> ImageType:
         pass

supervision/annotators/core.py

@@ -4,14 +4,20 @@ from typing import List, Optional, Tuple, Union
 import cv2
 import numpy as np
 
-from supervision.annotators.base import BaseAnnotator
-from supervision.annotators.utils import ColorLookup, Trace, resolve_color
+from supervision.annotators.base import BaseAnnotator, ImageType
+from supervision.annotators.utils import (
+    ColorLookup,
+    Trace,
+    resolve_color,
+    scene_to_annotator_img_type,
+)
 from supervision.config import CLASS_NAME_DATA_FIELD, ORIENTED_BOX_COORDINATES
 from supervision.detection.core import Detections
 from supervision.detection.utils import clip_boxes, mask_to_polygons
 from supervision.draw.color import Color, ColorPalette
 from supervision.draw.utils import draw_polygon
 from supervision.geometry.core import Position
+from supervision.utils.image import crop_image, place_image, resize_image
 
 
 class BoundingBoxAnnotator(BaseAnnotator):
@@ -37,23 +43,26 @@ class BoundingBoxAnnotator(BaseAnnotator):
         self.thickness: int = thickness
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with bounding boxes based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where bounding boxes will be drawn.
+            scene (ImageType): The image where bounding boxes will be drawn. `ImageType`
+            is a flexible type, accepting either `numpy.ndarray` or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -115,23 +124,27 @@ class OrientedBoxAnnotator(BaseAnnotator):
         self.thickness: int = thickness
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with oriented bounding boxes based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where bounding boxes will be drawn.
+            scene (ImageType): The image where bounding boxes will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -199,23 +212,27 @@ class MaskAnnotator(BaseAnnotator):
         self.opacity = opacity
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with masks based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where masks will be drawn.
+            scene (ImageType): The image where masks will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -282,23 +299,27 @@ class PolygonAnnotator(BaseAnnotator):
         self.thickness: int = thickness
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with polygons based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where polygons will be drawn.
+            scene (ImageType): The image where polygons will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -364,23 +385,27 @@ class ColorAnnotator(BaseAnnotator):
         self.color_lookup: ColorLookup = color_lookup
         self.opacity = opacity
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with box masks based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where bounding boxes will be drawn.
+            scene (ImageType): The image where bounding boxes will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -454,23 +479,27 @@ class HaloAnnotator(BaseAnnotator):
         self.color_lookup: ColorLookup = color_lookup
         self.kernel_size: int = kernel_size
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with halos based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where masks will be drawn.
+            scene (ImageType): The image where masks will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -548,23 +577,27 @@ class EllipseAnnotator(BaseAnnotator):
         self.end_angle: int = end_angle
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with ellipses based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where ellipses will be drawn.
+            scene (ImageType): The image where ellipses will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -635,23 +668,27 @@ class BoxCornerAnnotator(BaseAnnotator):
         self.corner_length: int = corner_length
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with box corners based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where box corners will be drawn.
+            scene (ImageType): The image where box corners will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -719,23 +756,27 @@ class CircleAnnotator(BaseAnnotator):
         self.thickness: int = thickness
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with circles based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where box corners will be drawn.
+            scene (ImageType): The image where box corners will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -805,23 +846,27 @@ class DotAnnotator(BaseAnnotator):
         self.position: Position = position
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with dots based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where dots will be drawn.
+            scene (ImageType): The image where dots will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -869,6 +914,7 @@ class LabelAnnotator:
         text_padding: int = 10,
         text_position: Position = Position.TOP_LEFT,
         color_lookup: ColorLookup = ColorLookup.CLASS,
+        border_radius: int = 0,
     ):
         """
         Args:
@@ -882,7 +928,10 @@ class LabelAnnotator:
                 Possible values are defined in the `Position` enum.
             color_lookup (str): Strategy for mapping colors to annotations.
                 Options are `INDEX`, `CLASS`, `TRACK`.
+            border_radius (int): The radius to apply round edges. If the selected
+                value is higher than the lower dimension, width or height, is clipped.
         """
+        self.border_radius: int = border_radius
         self.color: Union[Color, ColorPalette] = color
         self.text_color: Color = text_color
         self.text_scale: float = text_scale
@@ -929,38 +978,63 @@ class LabelAnnotator:
                 center_x + text_w // 2,
                 center_y + text_h,
             )
+        elif position == Position.CENTER_LEFT:
+            return (
+                center_x - text_w,
+                center_y - text_h // 2,
+                center_x,
+                center_y + text_h // 2,
+            )
+        elif position == Position.CENTER_RIGHT:
+            return (
+                center_x,
+                center_y - text_h // 2,
+                center_x + text_w,
+                center_y + text_h // 2,
+            )
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         labels: List[str] = None,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with labels based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where labels will be drawn.
+            scene (ImageType): The image where labels will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             labels (List[str]): Optional. Custom labels for each detection.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
-            import supervision as sv
+             import supervision as sv
 
             image = ...
             detections = sv.Detections(...)
 
+            labels = [
+                f"{class_name} {confidence:.2f}"
+                for class_name, confidence
+                in zip(detections['class_name'], detections.confidence)
+            ]
+
             label_annotator = sv.LabelAnnotator(text_position=sv.Position.CENTER)
             annotated_frame = label_annotator.annotate(
                 scene=image.copy(),
-                detections=detections
+                detections=detections,
+                labels=labels
             )
             ```
 
@@ -1017,12 +1091,11 @@ class LabelAnnotator:
             text_x = text_background_xyxy[0] + self.text_padding
             text_y = text_background_xyxy[1] + self.text_padding + text_h
 
-            cv2.rectangle(
-                img=scene,
-                pt1=(text_background_xyxy[0], text_background_xyxy[1]),
-                pt2=(text_background_xyxy[2], text_background_xyxy[3]),
+            self.draw_rounded_rectangle(
+                scene=scene,
+                xyxy=text_background_xyxy,
                 color=color.as_bgr(),
-                thickness=cv2.FILLED,
+                border_radius=self.border_radius,
             )
             cv2.putText(
                 img=scene,
@@ -1036,6 +1109,48 @@ class LabelAnnotator:
             )
         return scene
 
+    @staticmethod
+    def draw_rounded_rectangle(
+        scene: np.ndarray,
+        xyxy: Tuple[int, int, int, int],
+        color: Tuple[int, int, int],
+        border_radius: int,
+    ) -> np.ndarray:
+        x1, y1, x2, y2 = xyxy
+        width = x2 - x1
+        height = y2 - y1
+
+        border_radius = min(border_radius, min(width, height) // 2)
+
+        rectangle_coordinates = [
+            ((x1 + border_radius, y1), (x2 - border_radius, y2)),
+            ((x1, y1 + border_radius), (x2, y2 - border_radius)),
+        ]
+        circle_centers = [
+            (x1 + border_radius, y1 + border_radius),
+            (x2 - border_radius, y1 + border_radius),
+            (x1 + border_radius, y2 - border_radius),
+            (x2 - border_radius, y2 - border_radius),
+        ]
+
+        for coordinates in rectangle_coordinates:
+            cv2.rectangle(
+                img=scene,
+                pt1=coordinates[0],
+                pt2=coordinates[1],
+                color=color,
+                thickness=-1,
+            )
+        for center in circle_centers:
+            cv2.circle(
+                img=scene,
+                center=center,
+                radius=border_radius,
+                color=color,
+                thickness=-1,
+            )
+        return scene
+
 
 class BlurAnnotator(BaseAnnotator):
     """
@@ -1049,20 +1164,24 @@ class BlurAnnotator(BaseAnnotator):
         """
         self.kernel_size: int = kernel_size
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene by blurring regions based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where blurring will be applied.
+            scene (ImageType): The image where blurring will be applied.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -1130,24 +1249,28 @@ class TraceAnnotator:
         self.thickness = thickness
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Draws trace paths on the frame based on the detection coordinates provided.
 
         Args:
-            scene (np.ndarray): The image on which the traces will be drawn.
+            scene (ImageType): The image on which the traces will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): The detections which include coordinates for
                 which the traces will be drawn.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -1233,16 +1356,20 @@ class HeatMapAnnotator:
         self.top_hue = top_hue
         self.low_hue = low_hue
 
-    def annotate(self, scene: np.ndarray, detections: Detections) -> np.ndarray:
+    @scene_to_annotator_img_type
+    def annotate(self, scene: ImageType, detections: Detections) -> ImageType:
         """
         Annotates the scene with a heatmap based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where the heatmap will be drawn.
+            scene (ImageType): The image where the heatmap will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
 
         Returns:
-            Annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -1305,21 +1432,25 @@ class PixelateAnnotator(BaseAnnotator):
         """
         self.pixel_size: int = pixel_size
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene by pixelating regions based on the provided
             detections.
 
         Args:
-            scene (np.ndarray): The image where pixelating will be applied.
+            scene (ImageType): The image where pixelating will be applied.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -1389,23 +1520,27 @@ class TriangleAnnotator(BaseAnnotator):
         self.position: Position = position
         self.color_lookup: ColorLookup = color_lookup
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with triangles based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where triangles will be drawn.
+            scene (ImageType): The image where triangles will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            np.ndarray: The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -1481,24 +1616,28 @@ class RoundBoxAnnotator(BaseAnnotator):
             raise ValueError("roundness attribute must be float between (0, 1.0]")
         self.roundness: float = roundness
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with bounding boxes with rounded edges
         based on the provided detections.
 
         Args:
-            scene (np.ndarray): The image where rounded bounding boxes will be drawn.
+            scene (ImageType): The image where rounded bounding boxes will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -1614,20 +1753,23 @@ class PercentageBarAnnotator(BaseAnnotator):
         if border_thickness is None:
             self.border_thickness = int(0.15 * self.height)
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         custom_color_lookup: Optional[np.ndarray] = None,
         custom_values: Optional[np.ndarray] = None,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Annotates the given scene with percentage bars based on the provided
         detections. The percentage bars visually represent the confidence or custom
         values associated with each detection.
 
         Args:
-            scene (np.ndarray): The image where percentage bars will be drawn.
+            scene (ImageType): The image where percentage bars will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): Object detections to annotate.
             custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
                 Allows to override the default color mapping strategy.
@@ -1638,7 +1780,8 @@ class PercentageBarAnnotator(BaseAnnotator):
                 to be displayed.
 
         Returns:
-            The annotated image.
+            The annotated image, matching the type of `scene` (`numpy.ndarray`
+                or `PIL.Image.Image`)
 
         Example:
             ```python
@@ -1647,7 +1790,7 @@ class PercentageBarAnnotator(BaseAnnotator):
             image = ...
             detections = sv.Detections(...)
 
-            percentage_bar_annotator = sv.BoundingBoxAnnotator()
+            percentage_bar_annotator = sv.PercentageBarAnnotator()
             annotated_frame = percentage_bar_annotator.annotate(
                 scene=image.copy(),
                 detections=detections
@@ -1749,3 +1892,148 @@ class PercentageBarAnnotator(BaseAnnotator):
 
             if not all(0 <= value <= 1 for value in custom_values):
                 raise ValueError("All values in custom_values must be between 0 and 1.")
+
+
+class CropAnnotator(BaseAnnotator):
+    """
+    A class for drawing scaled up crops of detections on the scene.
+    """
+
+    def __init__(
+        self,
+        position: Position = Position.TOP_CENTER,
+        scale_factor: int = 2,
+        border_color: Union[Color, ColorPalette] = ColorPalette.DEFAULT,
+        border_thickness: int = 2,
+        border_color_lookup: ColorLookup = ColorLookup.CLASS,
+    ):
+        """
+        Args:
+            position (Position): The anchor position for placing the cropped and scaled
+                part of the detection in the scene.
+            scale_factor (int): The factor by which to scale the cropped image part. A
+                factor of 2, for example, would double the size of the cropped area,
+                allowing for a closer view of the detection.
+            border_color (Union[Color, ColorPalette]): The color or color palette to
+                use for annotating border around the cropped area.
+            border_thickness (int): The thickness of the border around the cropped area.
+            border_color_lookup (ColorLookup): Strategy for mapping colors to
+                annotations. Options are `INDEX`, `CLASS`, `TRACK`.
+        """
+        self.position: Position = position
+        self.scale_factor: int = scale_factor
+        self.border_color: Union[Color, ColorPalette] = border_color
+        self.border_thickness: int = border_thickness
+        self.border_color_lookup: ColorLookup = border_color_lookup
+
+    @scene_to_annotator_img_type
+    def annotate(
+        self,
+        scene: ImageType,
+        detections: Detections,
+        custom_color_lookup: Optional[np.ndarray] = None,
+    ) -> ImageType:
+        """
+        Annotates the provided scene with scaled and cropped parts of the image based
+        on the provided detections. Each detection is cropped from the original scene
+        and scaled according to the annotator's scale factor before being placed back
+        onto the scene at the specified position.
+
+
+        Args:
+            scene (ImageType): The image where cropped detection will be placed.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
+            detections (Detections): Object detections to annotate.
+            custom_color_lookup (Optional[np.ndarray]): Custom color lookup array.
+                Allows to override the default color mapping strategy.
+
+        Returns:
+            The annotated image.
+
+        Example:
+            ```python
+            import supervision as sv
+
+            image = ...
+            detections = sv.Detections(...)
+
+            crop_annotator = sv.CropAnnotator()
+            annotated_frame = crop_annotator.annotate(
+                scene=image.copy(),
+                detections=detections
+            )
+            ```
+        """
+        crops = [
+            crop_image(image=scene, xyxy=xyxy) for xyxy in detections.xyxy.astype(int)
+        ]
+        resized_crops = [
+            resize_image(image=crop, scale_factor=self.scale_factor) for crop in crops
+        ]
+        anchors = detections.get_anchors_coordinates(anchor=self.position).astype(int)
+
+        for idx, (resized_crop, anchor) in enumerate(zip(resized_crops, anchors)):
+            crop_wh = resized_crop.shape[1], resized_crop.shape[0]
+            (x1, y1), (x2, y2) = self.calculate_crop_coordinates(
+                anchor=anchor, crop_wh=crop_wh, position=self.position
+            )
+            scene = place_image(scene=scene, image=resized_crop, anchor=(x1, y1))
+            color = resolve_color(
+                color=self.border_color,
+                detections=detections,
+                detection_idx=idx,
+                color_lookup=self.border_color_lookup
+                if custom_color_lookup is None
+                else custom_color_lookup,
+            )
+            cv2.rectangle(
+                img=scene,
+                pt1=(x1, y1),
+                pt2=(x2, y2),
+                color=color.as_bgr(),
+                thickness=self.border_thickness,
+            )
+
+        return scene
+
+    @staticmethod
+    def calculate_crop_coordinates(
+        anchor: Tuple[int, int], crop_wh: Tuple[int, int], position: Position
+    ) -> Tuple[Tuple[int, int], Tuple[int, int]]:
+        anchor_x, anchor_y = anchor
+        width, height = crop_wh
+
+        if position == Position.TOP_LEFT:
+            return (anchor_x - width, anchor_y - height), (anchor_x, anchor_y)
+        elif position == Position.TOP_CENTER:
+            return (
+                (anchor_x - width // 2, anchor_y - height),
+                (anchor_x + width // 2, anchor_y),
+            )
+        elif position == Position.TOP_RIGHT:
+            return (anchor_x, anchor_y - height), (anchor_x + width, anchor_y)
+        elif position == Position.CENTER_LEFT:
+            return (
+                (anchor_x - width, anchor_y - height // 2),
+                (anchor_x, anchor_y + height // 2),
+            )
+        elif position == Position.CENTER or position == Position.CENTER_OF_MASS:
+            return (
+                (anchor_x - width // 2, anchor_y - height // 2),
+                (anchor_x + width // 2, anchor_y + height // 2),
+            )
+        elif position == Position.CENTER_RIGHT:
+            return (
+                (anchor_x, anchor_y - height // 2),
+                (anchor_x + width, anchor_y + height // 2),
+            )
+        elif position == Position.BOTTOM_LEFT:
+            return (anchor_x - width, anchor_y), (anchor_x, anchor_y + height)
+        elif position == Position.BOTTOM_CENTER:
+            return (
+                (anchor_x - width // 2, anchor_y),
+                (anchor_x + width // 2, anchor_y + height),
+            )
+        elif position == Position.BOTTOM_RIGHT:
+            return (anchor_x, anchor_y), (anchor_x + width, anchor_y + height)

supervision/annotators/utils.py

@@ -1,8 +1,12 @@
 from enum import Enum
+from functools import wraps
 from typing import Optional, Union
 
+import cv2
 import numpy as np
+from PIL import Image
 
+from supervision.annotators.base import ImageType
 from supervision.detection.core import Detections
 from supervision.draw.color import Color, ColorPalette
 from supervision.geometry.core import Position
@@ -119,3 +123,33 @@ class Trace:
 
     def get(self, tracker_id: int) -> np.ndarray:
         return self.xy[self.tracker_id == tracker_id]
+
+
+def pillow_to_cv2(image: Image.Image) -> np.ndarray:
+    scene = np.array(image)
+    scene = cv2.cvtColor(scene, cv2.COLOR_RGB2BGR)
+    return scene
+
+
+def scene_to_annotator_img_type(annotate_func):
+    """
+    Decorates `BaseAnnotator.annotate` implementations, converts scene to
+    an image type used internally by the annotators, converts back when annotation
+    is complete.
+    """
+
+    @wraps(annotate_func)
+    def wrapper(self, scene: ImageType, *args, **kwargs):
+        if isinstance(scene, np.ndarray):
+            return annotate_func(self, scene, *args, **kwargs)
+
+        if isinstance(scene, Image.Image):
+            scene = pillow_to_cv2(scene)
+            annotated = annotate_func(self, scene, *args, **kwargs)
+            annotated = cv2.cvtColor(annotated, cv2.COLOR_BGR2RGB)
+            annotated = Image.fromarray(annotated)
+            return annotated
+
+        raise ValueError(f"Unsupported image type: {type(scene)}")
+
+    return wrapper

supervision/detection/annotate.py

@@ -1,8 +1,9 @@
 from typing import List, Optional, Union
 
 import cv2
-import numpy as np
 
+from supervision.annotators.base import ImageType
+from supervision.annotators.utils import scene_to_annotator_img_type
 from supervision.detection.core import Detections
 from supervision.draw.color import Color, ColorPalette
 from supervision.utils.internal import deprecated
@@ -45,18 +46,21 @@ class BoxAnnotator:
         self.text_thickness: int = text_thickness
         self.text_padding: int = text_padding
 
+    @scene_to_annotator_img_type
     def annotate(
         self,
-        scene: np.ndarray,
+        scene: ImageType,
         detections: Detections,
         labels: Optional[List[str]] = None,
         skip_label: bool = False,
-    ) -> np.ndarray:
+    ) -> ImageType:
         """
         Draws bounding boxes on the frame using the detections provided.
 
         Args:
-            scene (np.ndarray): The image on which the bounding boxes will be drawn
+            scene (ImageType): The image on which the bounding boxes will be drawn.
+                `ImageType` is a flexible type, accepting either `numpy.ndarray`
+                or `PIL.Image.Image`.
             detections (Detections): The detections for which the
                 bounding boxes will be drawn
             labels (Optional[List[str]]): An optional list of labels
@@ -64,7 +68,8 @@ class BoxAnnotator:
                 corresponding `class_id` will be used as label.
             skip_label (bool): Is set to `True`, skips bounding box label annotation.
         Returns:
-            np.ndarray: The image with the bounding boxes drawn on it
+            ImageType: The image with the bounding boxes drawn on it, matching the
+                type of `scene` (`numpy.ndarray` or `PIL.Image.Image`)
 
         Example:
             ```python

supervision/detection/core.py

@@ -8,12 +8,14 @@ import numpy as np
 
 from supervision.config import CLASS_NAME_DATA_FIELD, ORIENTED_BOX_COORDINATES
 from supervision.detection.utils import (
+    box_non_max_suppression,
     calculate_masks_centroids,
     extract_ultralytics_masks,
     get_data_item,
     is_data_equal,
+    mask_non_max_suppression,
+    mask_to_xyxy,
     merge_data,
-    non_max_suppression,
     process_roboflow_result,
     validate_detections_fields,
     xywh_to_xyxy,
@@ -389,18 +391,30 @@ class Detections:
     @classmethod
     def from_transformers(cls, transformers_results: dict) -> Detections:
         """
-        Creates a Detections instance from object detection
+        Creates a Detections instance from object detection or segmentation
         [transformer](https://github.com/huggingface/transformers) inference result.
 
         Returns:
             Detections: A new Detections object.
         """
+        boxes = transformers_results.get("boxes")
 
-        return cls(
-            xyxy=transformers_results["boxes"].cpu().numpy(),
-            confidence=transformers_results["scores"].cpu().numpy(),
-            class_id=transformers_results["labels"].cpu().numpy().astype(int),
-        )
+        # If the boxes key is in the transformers_results then we know it's an
+        # object detection result. Else, we can assume it's a segmentation model
+        if boxes:
+            return cls(
+                xyxy=transformers_results["boxes"].cpu().numpy(),
+                confidence=transformers_results["scores"].cpu().numpy(),
+                class_id=transformers_results["labels"].cpu().numpy().astype(int),
+            )
+        else:
+            masks = transformers_results["masks"].cpu().numpy().astype(bool)
+            return cls(
+                xyxy=mask_to_xyxy(masks),
+                mask=masks,
+                confidence=transformers_results["scores"].cpu().numpy(),
+                class_id=transformers_results["labels"].cpu().numpy().astype(int),
+            )
 
     @classmethod
     def from_detectron2(cls, detectron2_results) -> Detections:
@@ -486,7 +500,9 @@ class Detections:
         )
 
         if np.asarray(xyxy).shape[0] == 0:
-            return cls.empty()
+            empty_detection = cls.empty()
+            empty_detection.data = {CLASS_NAME_DATA_FIELD: np.empty(0)}
+            return empty_detection
 
         return cls(
             xyxy=xyxy,
@@ -1001,7 +1017,8 @@ class Detections:
         self, threshold: float = 0.5, class_agnostic: bool = False
     ) -> Detections:
         """
-        Perform non-maximum suppression on the current set of object detections.
+        Performs non-max suppression on detection set. If the detections result
+        from a segmentation model, the IoU mask is applied. Otherwise, box IoU is used.
 
         Args:
             threshold (float, optional): The intersection-over-union threshold
@@ -1028,18 +1045,26 @@ class Detections:
 
         if class_agnostic:
             predictions = np.hstack((self.xyxy, self.confidence.reshape(-1, 1)))
-            indices = non_max_suppression(
+        else:
+            assert self.class_id is not None, (
+                "Detections class_id must be given for NMS to be executed. If you"
+                " intended to perform class agnostic NMS set class_agnostic=True."
+            )
+            predictions = np.hstack(
+                (
+                    self.xyxy,
+                    self.confidence.reshape(-1, 1),
+                    self.class_id.reshape(-1, 1),
+                )
+            )
+
+        if self.mask is not None:
+            indices = mask_non_max_suppression(
+                predictions=predictions, masks=self.mask, iou_threshold=threshold
+            )
+        else:
+            indices = box_non_max_suppression(
                 predictions=predictions, iou_threshold=threshold
             )
-            return self[indices]
 
-        assert self.class_id is not None, (
-            "Detections class_id must be given for NMS to be executed. If you intended"
-            " to perform class agnostic NMS set class_agnostic=True."
-        )
-
-        predictions = np.hstack(
-            (self.xyxy, self.confidence.reshape(-1, 1), self.class_id.reshape(-1, 1))
-        )
-        indices = non_max_suppression(predictions=predictions, iou_threshold=threshold)
         return self[indices]

supervision/detection/line_zone.py

@@ -14,6 +14,12 @@ class LineZone:
     This class is responsible for counting the number of objects that cross a
     predefined line.
 
+    <video controls>
+        <source
+            src="https://media.roboflow.com/supervision/cookbooks/count-objects-crossing-the-line-result-1280x720.mp4"
+            type="video/mp4">
+    </video>
+
     !!! warning
 
         LineZone uses the `tracker_id`. Read
@@ -25,7 +31,28 @@ class LineZone:
             to inside.
         out_count (int): The number of objects that have crossed the line from inside
             to outside.
-    """
+
+    Example:
+        ```python
+        import supervision as sv
+        from ultralytics import YOLO
+
+        model = YOLO(<SOURCE_MODEL_PATH>)
+        tracker = sv.ByteTrack()
+        frames_generator = sv.get_video_frames_generator(<SOURCE_VIDEO_PATH>)
+        start, end = sv.Point(x=0, y=1080), sv.Point(x=3840, y=1080)
+        line_zone = sv.LineZone(start=start, end=end)
+
+        for frame in frames_generator:
+            result = model(frame)[0]
+            detections = sv.Detections.from_ultralytics(result)
+            detections = tracker.update_with_detections(detections)
+            crossed_in, crossed_out = line_zone.trigger(detections)
+
+        line_zone.in_count, line_zone.out_count
+        # 7, 2
+        ```
+    """  # noqa: E501 // docs
 
     def __init__(
         self,

supervision/detection/tools/csv_sink.py

@@ -42,7 +42,7 @@ class CSVSink:
         csv_sink = sv.CSVSink(<RESULT_CSV_FILE_PATH>)
         frames_generator = sv.get_video_frames_generator(<SOURCE_VIDEO_PATH>)
 
-        with csv_sink:
+        with csv_sink as sink:
             for frame in frames_generator:
                 result = model(frame)[0]
                 detections = sv.Detections.from_ultralytics(result)

supervision/detection/tools/json_sink.py

@@ -0,0 +1,142 @@
+from __future__ import annotations
+
+import json
+import os
+from typing import Any, Dict, List, Optional
+
+from supervision.detection.core import Detections
+
+
+class JSONSink:
+    """
+    A utility class for saving detection data to a JSON file. This class is designed to
+    efficiently serialize detection objects into a JSON format, allowing for the
+    inclusion of bounding box coordinates and additional attributes like `confidence`,
+    `class_id`, and `tracker_id`.
+
+    !!! tip
+
+        JSONsink allow to pass custom data alongside the detection fields, providing
+        flexibility for logging various types of information.
+
+    Args:
+        file_name (str): The name of the JSON file where the detections will be stored.
+            Defaults to 'output.json'.
+
+    Example:
+        ```python
+        import supervision as sv
+        from ultralytics import YOLO
+
+        model = YOLO(<SOURCE_MODEL_PATH>)
+        json_sink = sv.JSONSink(<RESULT_JSON_FILE_PATH>)
+        frames_generator = sv.get_video_frames_generator(<SOURCE_VIDEO_PATH>)
+
+        with json_sink as sink:
+            for frame in frames_generator:
+                result = model(frame)[0]
+                detections = sv.Detections.from_ultralytics(result)
+                sink.append(detections, custom_data={'<CUSTOM_LABEL>':'<CUSTOM_DATA>'})
+        ```
+    """  # noqa: E501 // docs
+
+    def __init__(self, file_name: str = "output.json") -> None:
+        """
+        Initialize the JSONSink instance.
+
+        Args:
+            file_name (str): The name of the JSON file.
+
+        Returns:
+            None
+        """
+        self.file_name = file_name
+        self.file: Optional[open] = None
+        self.data: List[Dict[str, Any]] = []
+
+    def __enter__(self) -> JSONSink:
+        self.open()
+        return self
+
+    def __exit__(
+        self,
+        exc_type: Optional[type],
+        exc_val: Optional[Exception],
+        exc_tb: Optional[Any],
+    ) -> None:
+        self.write_and_close()
+
+    def open(self) -> None:
+        """
+        Open the JSON file for writing.
+
+        Returns:
+            None
+        """
+        parent_directory = os.path.dirname(self.file_name)
+        if parent_directory and not os.path.exists(parent_directory):
+            os.makedirs(parent_directory)
+
+        self.file = open(self.file_name, "w")
+
+    def write_and_close(self) -> None:
+        """
+        Write and close the JSON file.
+
+        Returns:
+            None
+        """
+        if self.file:
+            json.dump(self.data, self.file, indent=4)
+            self.file.close()
+
+    @staticmethod
+    def parse_detection_data(
+        detections: Detections, custom_data: Dict[str, Any] = None
+    ) -> List[Dict[str, Any]]:
+        parsed_rows = []
+        for i in range(len(detections.xyxy)):
+            row = {
+                "x_min": float(detections.xyxy[i][0]),
+                "y_min": float(detections.xyxy[i][1]),
+                "x_max": float(detections.xyxy[i][2]),
+                "y_max": float(detections.xyxy[i][3]),
+                "class_id": ""
+                if detections.class_id is None
+                else int(detections.class_id[i]),
+                "confidence": ""
+                if detections.confidence is None
+                else float(detections.confidence[i]),
+                "tracker_id": ""
+                if detections.tracker_id is None
+                else int(detections.tracker_id[i]),
+            }
+
+            if hasattr(detections, "data"):
+                for key, value in detections.data.items():
+                    row[key] = (
+                        str(value[i])
+                        if hasattr(value, "__getitem__") and value.ndim != 0
+                        else str(value)
+                    )
+
+            if custom_data:
+                row.update(custom_data)
+            parsed_rows.append(row)
+        return parsed_rows
+
+    def append(
+        self, detections: Detections, custom_data: Dict[str, Any] = None
+    ) -> None:
+        """
+        Append detection data to the JSON file.
+
+        Args:
+            detections (Detections): The detection data.
+            custom_data (Dict[str, Any]): Custom data to include.
+
+        Returns:
+            None
+        """
+        parsed_rows = JSONSink.parse_detection_data(detections, custom_data)
+        self.data.extend(parsed_rows)

supervision/detection/tools/polygon_zone.py

@@ -1,8 +1,9 @@
 from dataclasses import replace
-from typing import Optional, Tuple
+from typing import Iterable, Optional, Tuple
 
 import cv2
 import numpy as np
+import numpy.typing as npt
 
 from supervision import Detections
 from supervision.detection.utils import clip_boxes, polygon_to_mask
@@ -10,6 +11,7 @@ from supervision.draw.color import Color
 from supervision.draw.utils import draw_polygon, draw_text
 from supervision.geometry.core import Position
 from supervision.geometry.utils import get_polygon_center
+from supervision.utils.internal import deprecated_parameter
 
 
 class PolygonZone:
@@ -20,21 +22,32 @@ class PolygonZone:
         polygon (np.ndarray): A polygon represented by a numpy array of shape
             `(N, 2)`, containing the `x`, `y` coordinates of the points.
         frame_resolution_wh (Tuple[int, int]): The frame resolution (width, height)
-        triggering_position (Position): The position within the bounding
-            box that triggers the zone (default: Position.BOTTOM_CENTER)
+        triggering_anchors (Iterable[sv.Position]): A list of positions specifying
+            which anchors of the detections bounding box to consider when deciding on
+            whether the detection fits within the PolygonZone
+            (default: (sv.Position.BOTTOM_CENTER,)).
         current_count (int): The current count of detected objects within the zone
         mask (np.ndarray): The 2D bool mask for the polygon zone
     """
 
+    @deprecated_parameter(
+        old_parameter="triggering_position",
+        new_parameter="triggering_anchors",
+        map_function=lambda x: [x],
+        warning_message="`{old_parameter}` in `{function_name}` is deprecated and will "
+        "be remove in `supervision-0.23.0`. Use '{new_parameter}' "
+        "instead.",
+    )
     def __init__(
         self,
-        polygon: np.ndarray,
+        polygon: npt.NDArray[np.int64],
         frame_resolution_wh: Tuple[int, int],
-        triggering_position: Position = Position.BOTTOM_CENTER,
+        triggering_anchors: Iterable[Position] = (Position.BOTTOM_CENTER,),
     ):
         self.polygon = polygon.astype(int)
         self.frame_resolution_wh = frame_resolution_wh
-        self.triggering_position = triggering_position
+        self.triggering_anchors = triggering_anchors
+
         self.current_count = 0
 
         width, height = frame_resolution_wh
@@ -42,7 +55,7 @@ class PolygonZone:
             polygon=polygon, resolution_wh=(width + 1, height + 1)
         )
 
-    def trigger(self, detections: Detections) -> np.ndarray:
+    def trigger(self, detections: Detections) -> npt.NDArray[np.bool_]:
         """
         Determines if the detections are within the polygon zone.
 
@@ -59,10 +72,20 @@ class PolygonZone:
             xyxy=detections.xyxy, resolution_wh=self.frame_resolution_wh
         )
         clipped_detections = replace(detections, xyxy=clipped_xyxy)
-        clipped_anchors = np.ceil(
-            clipped_detections.get_anchors_coordinates(anchor=self.triggering_position)
-        ).astype(int)
-        is_in_zone = self.mask[clipped_anchors[:, 1], clipped_anchors[:, 0]]
+        all_clipped_anchors = np.array(
+            [
+                np.ceil(clipped_detections.get_anchors_coordinates(anchor)).astype(int)
+                for anchor in self.triggering_anchors
+            ]
+        )
+
+        is_in_zone: npt.NDArray[np.bool_] = (
+            self.mask[all_clipped_anchors[:, :, 1], all_clipped_anchors[:, :, 0]]
+            .transpose()
+            .astype(bool)
+        )
+
+        is_in_zone: npt.NDArray[np.bool_] = np.all(is_in_zone, axis=1)
         self.current_count = int(np.sum(is_in_zone))
         return is_in_zone.astype(bool)
 
@@ -83,6 +106,7 @@ class PolygonZoneAnnotator:
         font (int): The font type for the text on the polygon,
             default is cv2.FONT_HERSHEY_SIMPLEX
         center (Tuple[int, int]): The center of the polygon for text placement
+        display_in_zone_count (bool): Show the label of the zone or not. Default is True
     """
 
     def __init__(
@@ -94,6 +118,7 @@ class PolygonZoneAnnotator:
         text_scale: float = 0.5,
         text_thickness: int = 1,
         text_padding: int = 10,
+        display_in_zone_count: bool = True,
     ):
         self.zone = zone
         self.color = color
@@ -104,6 +129,7 @@ class PolygonZoneAnnotator:
         self.text_padding = text_padding
         self.font = cv2.FONT_HERSHEY_SIMPLEX
         self.center = get_polygon_center(polygon=zone.polygon)
+        self.display_in_zone_count = display_in_zone_count
 
     def annotate(self, scene: np.ndarray, label: Optional[str] = None) -> np.ndarray:
         """
@@ -124,16 +150,17 @@ class PolygonZoneAnnotator:
             thickness=self.thickness,
         )
 
-        annotated_frame = draw_text(
-            scene=annotated_frame,
-            text=str(self.zone.current_count) if label is None else label,
-            text_anchor=self.center,
-            background_color=self.color,
-            text_color=self.text_color,
-            text_scale=self.text_scale,
-            text_thickness=self.text_thickness,
-            text_padding=self.text_padding,
-            text_font=self.font,
-        )
+        if self.display_in_zone_count:
+            annotated_frame = draw_text(
+                scene=annotated_frame,
+                text=str(self.zone.current_count) if label is None else label,
+                text_anchor=self.center,
+                background_color=self.color,
+                text_color=self.text_color,
+                text_scale=self.text_scale,
+                text_thickness=self.text_thickness,
+                text_padding=self.text_padding,
+                text_font=self.font,
+            )
 
         return annotated_frame

supervision/detection/tools/smoother.py

@@ -115,4 +115,8 @@ class DetectionsSmoother:
             if track is not None:
                 tracked_detections.append(track)
 
-        return Detections.merge(tracked_detections)
+        detections = Detections.merge(tracked_detections)
+        if len(detections) == 0:
+            detections.tracker_id = np.array([], dtype=int)
+
+        return detections

supervision/detection/utils.py

@@ -59,7 +59,168 @@ def box_iou_batch(boxes_true: np.ndarray, boxes_detection: np.ndarray) -> np.nda
     return area_inter / (area_true[:, None] + area_detection - area_inter)
 
 
-def non_max_suppression(
+def _mask_iou_batch_split(
+    masks_true: np.ndarray, masks_detection: np.ndarray
+) -> np.ndarray:
+    """
+    Internal function.
+    Compute Intersection over Union (IoU) of two sets of masks -
+        `masks_true` and `masks_detection`.
+
+    Args:
+        masks_true (np.ndarray): 3D `np.ndarray` representing ground-truth masks.
+        masks_detection (np.ndarray): 3D `np.ndarray` representing detection masks.
+
+    Returns:
+        np.ndarray: Pairwise IoU of masks from `masks_true` and `masks_detection`.
+    """
+    intersection_area = np.logical_and(masks_true[:, None], masks_detection).sum(
+        axis=(2, 3)
+    )
+
+    masks_true_area = masks_true.sum(axis=(1, 2))
+    masks_detection_area = masks_detection.sum(axis=(1, 2))
+    union_area = masks_true_area[:, None] + masks_detection_area - intersection_area
+
+    return np.divide(
+        intersection_area,
+        union_area,
+        out=np.zeros_like(intersection_area, dtype=float),
+        where=union_area != 0,
+    )
+
+
+def mask_iou_batch(
+    masks_true: np.ndarray,
+    masks_detection: np.ndarray,
+    memory_limit: int = 1024 * 5,
+) -> np.ndarray:
+    """
+    Compute Intersection over Union (IoU) of two sets of masks -
+        `masks_true` and `masks_detection`.
+
+    Args:
+        masks_true (np.ndarray): 3D `np.ndarray` representing ground-truth masks.
+        masks_detection (np.ndarray): 3D `np.ndarray` representing detection masks.
+        memory_limit (int, optional): memory limit in MB, default is 1024 * 5 MB (5GB).
+
+    Returns:
+        np.ndarray: Pairwise IoU of masks from `masks_true` and `masks_detection`.
+    """
+    memory = (
+        masks_true.shape[0]
+        * masks_true.shape[1]
+        * masks_true.shape[2]
+        * masks_detection.shape[0]
+        / 1024
+        / 1024
+    )
+    if memory <= memory_limit:
+        return _mask_iou_batch_split(masks_true, masks_detection)
+
+    ious = []
+    step = max(
+        memory_limit
+        * 1024
+        * 1024
+        // (
+            masks_detection.shape[0]
+            * masks_detection.shape[1]
+            * masks_detection.shape[2]
+        ),
+        1,
+    )
+    for i in range(0, masks_true.shape[0], step):
+        ious.append(_mask_iou_batch_split(masks_true[i : i + step], masks_detection))
+
+    return np.vstack(ious)
+
+
+def resize_masks(masks: np.ndarray, max_dimension: int = 640) -> np.ndarray:
+    """
+    Resize all masks in the array to have a maximum dimension of max_dimension,
+    maintaining aspect ratio.
+
+    Args:
+        masks (np.ndarray): 3D array of binary masks with shape (N, H, W).
+        max_dimension (int): The maximum dimension for the resized masks.
+
+    Returns:
+        np.ndarray: Array of resized masks.
+    """
+    max_height = np.max(masks.shape[1])
+    max_width = np.max(masks.shape[2])
+    scale = min(max_dimension / max_height, max_dimension / max_width)
+
+    new_height = int(scale * max_height)
+    new_width = int(scale * max_width)
+
+    x = np.linspace(0, max_width - 1, new_width).astype(int)
+    y = np.linspace(0, max_height - 1, new_height).astype(int)
+    xv, yv = np.meshgrid(x, y)
+
+    resized_masks = masks[:, yv, xv]
+
+    resized_masks = resized_masks.reshape(masks.shape[0], new_height, new_width)
+    return resized_masks
+
+
+def mask_non_max_suppression(
+    predictions: np.ndarray,
+    masks: np.ndarray,
+    iou_threshold: float = 0.5,
+    mask_dimension: int = 640,
+) -> np.ndarray:
+    """
+    Perform Non-Maximum Suppression (NMS) on segmentation predictions.
+
+    Args:
+        predictions (np.ndarray): A 2D array of object detection predictions in
+            the format of `(x_min, y_min, x_max, y_max, score)`
+            or `(x_min, y_min, x_max, y_max, score, class)`. Shape: `(N, 5)` or
+            `(N, 6)`, where N is the number of predictions.
+        masks (np.ndarray): A 3D array of binary masks corresponding to the predictions.
+            Shape: `(N, H, W)`, where N is the number of predictions, and H, W are the
+            dimensions of each mask.
+        iou_threshold (float, optional): The intersection-over-union threshold
+            to use for non-maximum suppression.
+        mask_dimension (int, optional): The dimension to which the masks should be
+            resized before computing IOU values. Defaults to 640.
+
+    Returns:
+        np.ndarray: A boolean array indicating which predictions to keep after
+            non-maximum suppression.
+
+    Raises:
+        AssertionError: If `iou_threshold` is not within the closed
+        range from `0` to `1`.
+    """
+    assert 0 <= iou_threshold <= 1, (
+        "Value of `iou_threshold` must be in the closed range from 0 to 1, "
+        f"{iou_threshold} given."
+    )
+    rows, columns = predictions.shape
+
+    if columns == 5:
+        predictions = np.c_[predictions, np.zeros(rows)]
+
+    sort_index = predictions[:, 4].argsort()[::-1]
+    predictions = predictions[sort_index]
+    masks = masks[sort_index]
+    masks_resized = resize_masks(masks, mask_dimension)
+    ious = mask_iou_batch(masks_resized, masks_resized)
+    categories = predictions[:, 5]
+
+    keep = np.ones(rows, dtype=bool)
+    for i in range(rows):
+        if keep[i]:
+            condition = (ious[i] > iou_threshold) & (categories[i] == categories)
+            keep[i + 1 :] = np.where(condition[i + 1 :], False, keep[i + 1 :])
+
+    return keep[sort_index.argsort()]
+
+
+def box_non_max_suppression(
     predictions: np.ndarray, iou_threshold: float = 0.5
 ) -> np.ndarray:
     """
@@ -684,8 +845,15 @@ def get_data_item(
         elif isinstance(value, list):
             if isinstance(index, slice):
                 subset_data[key] = value[index]
-            elif isinstance(index, (list, np.ndarray)):
+            elif isinstance(index, list):
                 subset_data[key] = [value[i] for i in index]
+            elif isinstance(index, np.ndarray):
+                if index.dtype == bool:
+                    subset_data[key] = [
+                        value[i] for i, index_value in enumerate(index) if index_value
+                    ]
+                else:
+                    subset_data[key] = [value[i] for i in index]
             elif isinstance(index, int):
                 subset_data[key] = [value[index]]
             else:

supervision/draw/color.py

@@ -176,31 +176,31 @@ class Color:
         return self.b, self.g, self.r
 
     @classproperty
-    def WHITE(cls):
+    def WHITE(cls) -> Color:
         return Color.from_hex("#FFFFFF")
 
     @classproperty
-    def BLACK(cls):
+    def BLACK(cls) -> Color:
         return Color.from_hex("#000000")
 
     @classproperty
-    def RED(cls):
+    def RED(cls) -> Color:
         return Color.from_hex("#FF0000")
 
     @classproperty
-    def GREEN(cls):
+    def GREEN(cls) -> Color:
         return Color.from_hex("#00FF00")
 
     @classproperty
-    def BLUE(cls):
+    def BLUE(cls) -> Color:
         return Color.from_hex("#0000FF")
 
     @classproperty
-    def YELLOW(cls):
+    def YELLOW(cls) -> Color:
         return Color.from_hex("#FFFF00")
 
     @classproperty
-    def ROBOFLOW(cls):
+    def ROBOFLOW(cls) -> Color:
         return Color.from_hex("#A351FB")
 
     @classmethod

supervision/draw/utils.py

@@ -238,13 +238,13 @@ def draw_image(
     return scene
 
 
-def calculate_dynamic_text_scale(resolution_wh: Tuple[int, int]) -> float:
+def calculate_optimal_text_scale(resolution_wh: Tuple[int, int]) -> float:
     """
-    Calculate a dynamic font scale based on the resolution of an image.
+    Calculate font scale based on the resolution of an image.
 
     Parameters:
          resolution_wh (Tuple[int, int]): A tuple representing the width and height
-                 of the image.
+             of the image.
 
     Returns:
          float: The calculated font scale factor.
@@ -252,25 +252,17 @@ def calculate_dynamic_text_scale(resolution_wh: Tuple[int, int]) -> float:
     return min(resolution_wh) * 1e-3
 
 
-def calculate_dynamic_line_thickness(resolution_wh: Tuple[int, int]) -> int:
+def calculate_optimal_line_thickness(resolution_wh: Tuple[int, int]) -> int:
     """
-    Calculate a dynamic line thickness based on the resolution of an image.
+    Calculate line thickness based on the resolution of an image.
 
     Parameters:
         resolution_wh (Tuple[int, int]): A tuple representing the width and height
-                of the image.
+            of the image.
 
     Returns:
         int: The calculated line thickness in pixels.
     """
-    min_dimension = min(resolution_wh)
-    if min_dimension < 480:
+    if min(resolution_wh) < 1080:
         return 2
-    if min_dimension < 720:
-        return 2
-    if min_dimension < 1080:
-        return 2
-    if min_dimension < 2160:
-        return 4
-    else:
-        return 4
+    return 4

supervision/metrics/detection.py

@@ -836,10 +836,10 @@ class MeanAveragePrecision:
             precision = true_positives / (true_positives + false_positives)
 
             for iou_level_idx in range(matches.shape[1]):
-                average_precisions[
-                    class_idx, iou_level_idx
-                ] = MeanAveragePrecision.compute_average_precision(
-                    recall[:, iou_level_idx], precision[:, iou_level_idx]
+                average_precisions[class_idx, iou_level_idx] = (
+                    MeanAveragePrecision.compute_average_precision(
+                        recall[:, iou_level_idx], precision[:, iou_level_idx]
+                    )
                 )
 
         return average_precisions

supervision/tracker/byte_tracker/core.py

@@ -3,9 +3,11 @@ from typing import List, Tuple
 import numpy as np
 
 from supervision.detection.core import Detections
+from supervision.detection.utils import box_iou_batch
 from supervision.tracker.byte_tracker import matching
 from supervision.tracker.byte_tracker.basetrack import BaseTrack, TrackState
 from supervision.tracker.byte_tracker.kalman_filter import KalmanFilter
+from supervision.utils.internal import deprecated_parameter
 
 
 class STrack(BaseTrack):
@@ -172,26 +174,57 @@ class ByteTrack:
     </video>
 
     Parameters:
-        track_thresh (float, optional): Detection confidence threshold
-            for track activation.
-        track_buffer (int, optional): Number of frames to buffer when a track is lost.
-        match_thresh (float, optional): Threshold for matching tracks with detections.
+        track_activation_threshold (float, optional): Detection confidence threshold
+            for track activation. Increasing track_activation_threshold improves accuracy
+            and stability but might miss true detections. Decreasing it increases
+            completeness but risks introducing noise and instability.
+        lost_track_buffer (int, optional): Number of frames to buffer when a track is lost.
+            Increasing lost_track_buffer enhances occlusion handling, significantly
+            reducing the likelihood of track fragmentation or disappearance caused
+            by brief detection gaps.
+        minimum_matching_threshold (float, optional): Threshold for matching tracks with detections.
+            Increasing minimum_matching_threshold improves accuracy but risks fragmentation.
+            Decreasing it improves completeness but risks false positives and drift.
         frame_rate (int, optional): The frame rate of the video.
     """  # noqa: E501 // docs
 
+    @deprecated_parameter(
+        old_parameter="track_buffer",
+        new_parameter="lost_track_buffer",
+        map_function=lambda x: x,
+        warning_message="`{old_parameter}` in `{function_name}` is deprecated and will "
+        "be remove in `supervision-0.23.0`. Use '{new_parameter}' "
+        "instead.",
+    )
+    @deprecated_parameter(
+        old_parameter="track_thresh",
+        new_parameter="track_activation_threshold",
+        map_function=lambda x: x,
+        warning_message="`{old_parameter}` in `{function_name}` is deprecated and will "
+        "be remove in `supervision-0.23.0`. Use '{new_parameter}' "
+        "instead.",
+    )
+    @deprecated_parameter(
+        old_parameter="match_thresh",
+        new_parameter="minimum_matching_threshold",
+        map_function=lambda x: x,
+        warning_message="`{old_parameter}` in `{function_name}` is deprecated and will "
+        "be remove in `supervision-0.23.0`. Use '{new_parameter}' "
+        "instead.",
+    )
     def __init__(
         self,
-        track_thresh: float = 0.25,
-        track_buffer: int = 30,
-        match_thresh: float = 0.8,
+        track_activation_threshold: float = 0.25,
+        lost_track_buffer: int = 30,
+        minimum_matching_threshold: float = 0.8,
         frame_rate: int = 30,
     ):
-        self.track_thresh = track_thresh
-        self.match_thresh = match_thresh
+        self.track_activation_threshold = track_activation_threshold
+        self.minimum_matching_threshold = minimum_matching_threshold
 
         self.frame_id = 0
-        self.det_thresh = self.track_thresh + 0.1
-        self.max_time_lost = int(frame_rate / 30.0 * track_buffer)
+        self.det_thresh = self.track_activation_threshold + 0.1
+        self.max_time_lost = int(frame_rate / 30.0 * lost_track_buffer)
         self.kalman_filter = KalmanFilter()
 
         self.tracked_tracks: List[STrack] = []
@@ -238,27 +271,28 @@ class ByteTrack:
             ```
         """
 
-        tracks = self.update_with_tensors(
-            tensors=detections2boxes(detections=detections)
-        )
-        detections = Detections.empty()
+        tensors = detections2boxes(detections=detections)
+        tracks = self.update_with_tensors(tensors=tensors)
+
         if len(tracks) > 0:
-            detections.xyxy = np.array(
-                [track.tlbr for track in tracks], dtype=np.float32
-            )
-            detections.class_id = np.array(
-                [int(t.class_ids) for t in tracks], dtype=int
-            )
-            detections.tracker_id = np.array(
-                [int(t.track_id) for t in tracks], dtype=int
-            )
-            detections.confidence = np.array(
-                [t.score for t in tracks], dtype=np.float32
-            )
+            detection_bounding_boxes = np.asarray([det[:4] for det in tensors])
+            track_bounding_boxes = np.asarray([track.tlbr for track in tracks])
+
+            ious = box_iou_batch(detection_bounding_boxes, track_bounding_boxes)
+
+            iou_costs = 1 - ious
+
+            matches, _, _ = matching.linear_assignment(iou_costs, 0.5)
+            detections.tracker_id = np.full(len(detections), -1, dtype=int)
+            for i_detection, i_track in matches:
+                detections.tracker_id[i_detection] = int(tracks[i_track].track_id)
+
+            return detections[detections.tracker_id != -1]
+
         else:
             detections.tracker_id = np.array([], dtype=int)
 
-        return detections
+            return detections
 
     def reset(self):
         """
@@ -295,9 +329,9 @@ class ByteTrack:
         scores = tensors[:, 4]
         bboxes = tensors[:, :4]
 
-        remain_inds = scores > self.track_thresh
+        remain_inds = scores > self.track_activation_threshold
         inds_low = scores > 0.1
-        inds_high = scores < self.track_thresh
+        inds_high = scores < self.track_activation_threshold
 
         inds_second = np.logical_and(inds_low, inds_high)
         dets_second = bboxes[inds_second]
@@ -335,7 +369,7 @@ class ByteTrack:
 
         dists = matching.fuse_score(dists, detections)
         matches, u_track, u_detection = matching.linear_assignment(
-            dists, thresh=self.match_thresh
+            dists, thresh=self.minimum_matching_threshold
         )
 
         for itracked, idet in matches:

supervision/utils/image.py

@@ -1,6 +1,6 @@
 import os
 import shutil
-from typing import Optional
+from typing import Optional, Tuple
 
 import cv2
 import numpy as np
@@ -32,8 +32,76 @@ def crop_image(image: np.ndarray, xyxy: np.ndarray) -> np.ndarray:
 
     xyxy = np.round(xyxy).astype(int)
     x1, y1, x2, y2 = xyxy
-    cropped_img = image[y1:y2, x1:x2]
-    return cropped_img
+    return image[y1:y2, x1:x2]
+
+
+def resize_image(image: np.ndarray, scale_factor: float) -> np.ndarray:
+    """
+    Resizes an image by a given scale factor using cv2.INTER_LINEAR interpolation.
+
+    Args:
+        image (np.ndarray): The input image to be resized.
+        scale_factor (float): The factor by which the image will be scaled. Scale factor
+            > 1.0 zooms in, < 1.0 zooms out.
+
+    Returns:
+        np.ndarray: The resized image.
+
+    Raises:
+        ValueError: If the scale factor is non-positive.
+    """
+    if scale_factor <= 0:
+        raise ValueError("Scale factor must be positive.")
+
+    old_width, old_height = image.shape[1], image.shape[0]
+    nwe_width = int(old_width * scale_factor)
+    new_height = int(old_height * scale_factor)
+
+    return cv2.resize(image, (nwe_width, new_height), interpolation=cv2.INTER_LINEAR)
+
+
+def place_image(
+    scene: np.ndarray, image: np.ndarray, anchor: Tuple[int, int]
+) -> np.ndarray:
+    """
+    Places an image onto a scene at a given anchor point, handling cases where
+    the image's position is partially or completely outside the scene's bounds.
+
+    Args:
+        scene (np.ndarray): The background scene onto which the image is placed.
+        image (np.ndarray): The image to be placed onto the scene.
+        anchor (Tuple[int, int]): The (x, y) coordinates in the scene where the
+            top-left corner of the image will be placed.
+
+    Returns:
+        np.ndarray: The modified scene with the image placed at the anchor point,
+            or unchanged if the image placement is completely outside the scene.
+    """
+    scene_height, scene_width = scene.shape[:2]
+    image_height, image_width = image.shape[:2]
+    anchor_x, anchor_y = anchor
+
+    is_out_horizontally = anchor_x + image_width <= 0 or anchor_x >= scene_width
+    is_out_vertically = anchor_y + image_height <= 0 or anchor_y >= scene_height
+
+    if is_out_horizontally or is_out_vertically:
+        return scene
+
+    start_y = max(anchor_y, 0)
+    start_x = max(anchor_x, 0)
+    end_y = min(scene_height, anchor_y + image_height)
+    end_x = min(scene_width, anchor_x + image_width)
+
+    crop_start_y = max(-anchor_y, 0)
+    crop_start_x = max(-anchor_x, 0)
+    crop_end_y = image_height - max((anchor_y + image_height) - scene_height, 0)
+    crop_end_x = image_width - max((anchor_x + image_width) - scene_width, 0)
+
+    scene[start_y:end_y, start_x:end_x] = image[
+        crop_start_y:crop_end_y, crop_start_x:crop_end_x
+    ]
+
+    return scene
 
 
 class ImageSink:
@@ -85,7 +153,8 @@ class ImageSink:
         Save a given image in the target directory.
 
         Args:
-            image (np.ndarray): The image to be saved.
+            image (np.ndarray): The image to be saved. The image must be in BGR color
+                format.
             image_name (str, optional): The name to use for the saved image.
                 If not provided, a name will be
                 generated using the `image_name_pattern`.

supervision/utils/internal.py

@@ -1,5 +1,104 @@
 import functools
+import os
 import warnings
+from typing import Callable
+
+
+class SupervisionWarnings(Warning):
+    """Supervision warning category.
+    Set the deprecation warnings visibility for Supervision library.
+    You can set the environment variable SUPERVISON_DEPRECATION_WARNING to '0' to
+    disable the deprecation warnings.
+    """
+
+    pass
+
+
+def format_warning(msg, category, filename, lineno, line=None):
+    """
+    Format a warning the same way as the default formatter, but also include the
+    category name in the output.
+    """
+    return f"{category.__name__}: {msg}\n"
+
+
+warnings.formatwarning = format_warning
+
+if os.getenv("SUPERVISON_DEPRECATION_WARNING") == "0":
+    warnings.simplefilter("ignore", SupervisionWarnings)
+else:
+    warnings.simplefilter("always", SupervisionWarnings)
+
+
+def deprecated_parameter(
+    old_parameter: str,
+    new_parameter: str,
+    map_function: Callable = lambda x: x,
+    warning_message: str = "Warning: '{old_parameter}' in '{function_name}' is "
+    "deprecated: use '{new_parameter}' instead.",
+    **message_kwargs,
+):
+    """
+    A decorator to mark a function's parameter as deprecated and issue a warning when
+    used.
+
+    Parameters:
+        old_parameter (str): The name of the deprecated parameter.
+        new_parameter (str): The name of the parameter that should be used instead.
+        map_function (Callable, optional): A function used to map the value of the old
+            parameter to the new parameter. Defaults to the identity function.
+        warning_message (str, optional): The warning message to be displayed when the
+            deprecated parameter is used. Defaults to a generic warning message with
+            placeholders for the old parameter, new parameter, and function name.
+        **message_kwargs: Additional keyword arguments that can be used to customize
+            the warning message.
+
+    Returns:
+        Callable: A decorator function that can be applied to mark a function's
+            parameter as deprecated.
+
+    Examples:
+        ```python
+        @deprecated_parameter(
+            old_parameter=<OLD_PARAMETER_NAME>,
+            new_parameter=<NEW_PARAMETER_NAME>
+        )
+        def example_function(<NEW_PARAMETER_NAME>):
+            pass
+
+        # call function using deprecated parameter
+        example_function(<OLD_PARAMETER_NAME>=<OLD_PARAMETER_VALUE>)
+        ```
+    """
+
+    def decorator(func):
+        @functools.wraps(func)
+        def wrapper(*args, **kwargs):
+            if old_parameter in kwargs:
+                if args and hasattr(args[0], "__class__"):
+                    class_name = args[0].__class__.__name__
+                    function_name = f"{class_name}.{func.__name__}"
+                else:
+                    function_name = func.__name__
+
+                warnings.warn(
+                    message=warning_message.format(
+                        function_name=function_name,
+                        old_parameter=old_parameter,
+                        new_parameter=new_parameter,
+                        **message_kwargs,
+                    ),
+                    category=SupervisionWarnings,
+                    stacklevel=2,
+                )
+
+                kwargs[new_parameter] = map_function(kwargs.pop(old_parameter))
+
+            return func(*args, **kwargs)
+
+        return wrapper
+
+    return decorator
 
 
 def deprecated(reason: str):
@@ -8,7 +107,7 @@ def deprecated(reason: str):
         def wrapper(*args, **kwargs):
             warnings.warn(
                 f"{func.__name__} is deprecated: {reason}",
-                category=DeprecationWarning,
+                category=SupervisionWarnings,
                 stacklevel=2,
             )
             return func(*args, **kwargs)

supervision/utils/notebook.py

@@ -2,17 +2,21 @@ from typing import List, Optional, Tuple
 
 import cv2
 import matplotlib.pyplot as plt
-import numpy as np
+from PIL import Image
+
+from supervision.annotators.base import ImageType
+from supervision.annotators.utils import pillow_to_cv2
 
 
 def plot_image(
-    image: np.ndarray, size: Tuple[int, int] = (12, 12), cmap: Optional[str] = "gray"
+    image: ImageType, size: Tuple[int, int] = (12, 12), cmap: Optional[str] = "gray"
 ) -> None:
     """
     Plots image using matplotlib.
 
     Args:
-        image (np.ndarray): The frame to be displayed.
+        image (ImageType): The frame to be displayed ImageType
+             is a flexible type, accepting either `numpy.ndarray` or `PIL.Image.Image`.
         size (Tuple[int, int]): The size of the plot.
         cmap (str): the colormap to use for single channel images.
 
@@ -27,6 +31,9 @@ def plot_image(
         sv.plot_image(image=image, size=(16, 16))
         ```
     """
+    if isinstance(image, Image.Image):
+        image = pillow_to_cv2(image)
+
     plt.figure(figsize=size)
 
     if image.ndim == 2:
@@ -39,7 +46,7 @@ def plot_image(
 
 
 def plot_images_grid(
-    images: List[np.ndarray],
+    images: List[ImageType],
     grid_size: Tuple[int, int],
     titles: Optional[List[str]] = None,
     size: Tuple[int, int] = (12, 12),
@@ -49,7 +56,8 @@ def plot_images_grid(
     Plots images in a grid using matplotlib.
 
     Args:
-       images (List[np.ndarray]): A list of images as numpy arrays.
+       images (List[ImageType]): A list of images as ImageType
+             is a flexible type, accepting either `numpy.ndarray` or `PIL.Image.Image`.
        grid_size (Tuple[int, int]): A tuple specifying the number
             of rows and columns for the grid.
        titles (Optional[List[str]]): A list of titles for each image.
@@ -65,9 +73,10 @@ def plot_images_grid(
         ```python
         import cv2
         import supervision as sv
+        from PIL import Image
 
         image1 = cv2.imread("path/to/image1.jpg")
-        image2 = cv2.imread("path/to/image2.jpg")
+        image2 = Image.open("path/to/image2.jpg")
         image3 = cv2.imread("path/to/image3.jpg")
 
         images = [image1, image2, image3]
@@ -79,6 +88,10 @@ def plot_images_grid(
     """
     nrows, ncols = grid_size
 
+    for idx, img in enumerate(images):
+        if isinstance(img, Image.Image):
+            images[idx] = pillow_to_cv2(img)
+
     if len(images) > nrows * ncols:
         raise ValueError(
             "The number of images exceeds the grid size. Please increase the grid size"

supervision/utils/video.py

@@ -28,7 +28,7 @@ class VideoInfo:
         ```python
         import supervision as sv
 
-        video_info = sv.VideoInfo.from_video_path(video_path='video.mp4')
+        video_info = sv.VideoInfo.from_video_path(video_path=<SOURCE_VIDEO_FILE>)
 
         video_info
         # VideoInfo(width=3840, height=2160, fps=25, total_frames=538)
@@ -75,14 +75,14 @@ class VideoSink:
         ```python
         import supervision as sv
 
-        video_info = sv.VideoInfo.from_video_path('source.mp4')
-        frames_generator = sv.get_video_frames_generator('source.mp4')
+        video_info = sv.VideoInfo.from_video_path(<SOURCE_VIDEO_PATH>)
+        frames_generator = sv.get_video_frames_generator(<SOURCE_VIDEO_PATH>)
 
-        with sv.VideoSink(target_path='target.mp4', video_info=video_info) as sink:
+        with sv.VideoSink(target_path=<TARGET_VIDEO_PATH>, video_info=video_info) as sink:
             for frame in frames_generator:
                 sink.write_frame(frame=frame)
         ```
-    """
+    """  # noqa: E501 // docs
 
     def __init__(self, target_path: str, video_info: VideoInfo, codec: str = "mp4v"):
         self.target_path = target_path
@@ -105,6 +105,13 @@ class VideoSink:
         return self
 
     def write_frame(self, frame: np.ndarray):
+        """
+        Writes a single video frame to the target video file.
+
+        Args:
+            frame (np.ndarray): The video frame to be written to the file. The frame
+                must be in BGR color format.
+        """
         self.__writer.write(frame)
 
     def __exit__(self, exc_type, exc_value, exc_traceback):
@@ -147,7 +154,7 @@ def get_video_frames_generator(
         ```python
         import supervision as sv
 
-        for frame in sv.get_video_frames_generator(source_path='source_video.mp4'):
+        for frame in sv.get_video_frames_generator(source_path=<SOURCE_VIDEO_PATH>):
             ...
         ```
     """
@@ -191,8 +198,8 @@ def process_video(
             ...
 
         process_video(
-            source_path='...',
-            target_path='...',
+            source_path=<SOURCE_VIDEO_PATH>,
+            target_path=<TARGET_VIDEO_PATH>,
             callback=callback
         )
         ```

test/detection/test_json.py

@@ -0,0 +1,249 @@
+import json
+import os
+from test.test_utils import mock_detections
+from typing import Any, Dict, List
+
+import pytest
+
+import supervision as sv
+
+
+@pytest.mark.parametrize(
+    "detections, custom_data, "
+    "second_detections, second_custom_data, "
+    "file_name, expected_result",
+    [
+        (
+            mock_detections(
+                xyxy=[[10, 20, 30, 40], [50, 60, 70, 80]],
+                confidence=[0.7, 0.8],
+                class_id=[0, 0],
+                tracker_id=[0, 1],
+                data={"class_name": ["person", "person"]},
+            ),
+            {"frame_number": 42},
+            mock_detections(
+                xyxy=[[15, 25, 35, 45], [55, 65, 75, 85]],
+                confidence=[0.6, 0.9],
+                class_id=[1, 1],
+                tracker_id=[2, 3],
+                data={"class_name": ["car", "car"]},
+            ),
+            {"frame_number": 43},
+            "test_detections.json",
+            [
+                {
+                    "x_min": 10,
+                    "y_min": 20,
+                    "x_max": 30,
+                    "y_max": 40,
+                    "class_id": 0,
+                    "confidence": 0.699999988079071,
+                    "tracker_id": 0,
+                    "class_name": "person",
+                    "frame_number": 42,
+                },
+                {
+                    "x_min": 50,
+                    "y_min": 60,
+                    "x_max": 70,
+                    "y_max": 80,
+                    "class_id": 0,
+                    "confidence": 0.800000011920929,
+                    "tracker_id": 1,
+                    "class_name": "person",
+                    "frame_number": 42,
+                },
+                {
+                    "x_min": 15,
+                    "y_min": 25,
+                    "x_max": 35,
+                    "y_max": 45,
+                    "class_id": 1,
+                    "confidence": 0.6000000238418579,
+                    "tracker_id": 2,
+                    "class_name": "car",
+                    "frame_number": 43,
+                },
+                {
+                    "x_min": 55,
+                    "y_min": 65,
+                    "x_max": 75,
+                    "y_max": 85,
+                    "class_id": 1,
+                    "confidence": 0.8999999761581421,
+                    "tracker_id": 3,
+                    "class_name": "car",
+                    "frame_number": 43,
+                },
+            ],
+        ),  # Multiple detections
+        (
+            mock_detections(
+                xyxy=[[60, 70, 80, 90], [100, 110, 120, 130]],
+                tracker_id=[4, 5],
+                data={"class_name": ["bike", "dog"]},
+            ),
+            {"frame_number": 44},
+            mock_detections(
+                xyxy=[[65, 75, 85, 95], [105, 115, 125, 135]],
+                confidence=[0.5, 0.4],
+                data={"class_name": ["tree", "cat"]},
+            ),
+            {"frame_number": 45},
+            "test_detections_missing_fields.json",
+            [
+                {
+                    "x_min": 60,
+                    "y_min": 70,
+                    "x_max": 80,
+                    "y_max": 90,
+                    "class_id": "",
+                    "confidence": "",
+                    "tracker_id": 4,
+                    "class_name": "bike",
+                    "frame_number": 44,
+                },
+                {
+                    "x_min": 100,
+                    "y_min": 110,
+                    "x_max": 120,
+                    "y_max": 130,
+                    "class_id": "",
+                    "confidence": "",
+                    "tracker_id": 5,
+                    "class_name": "dog",
+                    "frame_number": 44,
+                },
+                {
+                    "x_min": 65,
+                    "y_min": 75,
+                    "x_max": 85,
+                    "y_max": 95,
+                    "class_id": "",
+                    "confidence": 0.5,
+                    "tracker_id": "",
+                    "class_name": "tree",
+                    "frame_number": 45,
+                },
+                {
+                    "x_min": 105,
+                    "y_min": 115,
+                    "x_max": 125,
+                    "y_max": 135,
+                    "class_id": "",
+                    "confidence": 0.4000000059604645,
+                    "tracker_id": "",
+                    "class_name": "cat",
+                    "frame_number": 45,
+                },
+            ],
+        ),  # Missing fields
+        (
+            mock_detections(
+                xyxy=[[10, 11, 12, 13]],
+                confidence=[0.95],
+                data={"class_name": "unknown", "is_detected": True, "score": 1},
+            ),
+            {"frame_number": 46},
+            mock_detections(
+                xyxy=[[14, 15, 16, 17]],
+                data={"class_name": "artifact", "is_detected": False, "score": 0.85},
+            ),
+            {"frame_number": 47},
+            "test_detections_varied_data.json",
+            [
+                {
+                    "x_min": 10,
+                    "y_min": 11,
+                    "x_max": 12,
+                    "y_max": 13,
+                    "class_id": "",
+                    "confidence": 0.949999988079071,
+                    "tracker_id": "",
+                    "class_name": "unknown",
+                    "is_detected": "True",
+                    "score": "1",
+                    "frame_number": 46,
+                },
+                {
+                    "x_min": 14,
+                    "y_min": 15,
+                    "x_max": 16,
+                    "y_max": 17,
+                    "class_id": "",
+                    "confidence": "",
+                    "tracker_id": "",
+                    "class_name": "artifact",
+                    "is_detected": "False",
+                    "score": "0.85",
+                    "frame_number": 47,
+                },
+            ],
+        ),  # Inconsistent Data Types
+        (
+            mock_detections(
+                xyxy=[[20, 21, 22, 23]],
+            ),
+            {
+                "metadata": {"sensor_id": 101, "location": "north"},
+                "tags": ["urgent", "review"],
+            },
+            mock_detections(
+                xyxy=[[14, 15, 16, 17]],
+            ),
+            {
+                "metadata": {"sensor_id": 104, "location": "west"},
+                "tags": ["not-urgent", "done"],
+            },
+            "test_detections_complex_data.json",
+            [
+                {
+                    "x_min": 20,
+                    "y_min": 21,
+                    "x_max": 22,
+                    "y_max": 23,
+                    "class_id": "",
+                    "confidence": "",
+                    "tracker_id": "",
+                    "metadata": {"sensor_id": 101, "location": "north"},
+                    "tags": ["urgent", "review"],
+                },
+                {
+                    "x_min": 14,
+                    "y_min": 15,
+                    "x_max": 16,
+                    "y_max": 17,
+                    "class_id": "",
+                    "confidence": "",
+                    "tracker_id": "",
+                    "metadata": {"sensor_id": 104, "location": "west"},
+                    "tags": ["not-urgent", "done"],
+                },
+            ],
+        ),  # Complex Data
+    ],
+)
+def test_json_sink(
+    detections: mock_detections,
+    custom_data: Dict[str, Any],
+    second_detections: mock_detections,
+    second_custom_data: Dict[str, Any],
+    file_name: str,
+    expected_result: List[List[Any]],
+) -> None:
+    with sv.JSONSink(file_name) as sink:
+        sink.append(detections, custom_data)
+        sink.append(second_detections, second_custom_data)
+
+    assert_json_equal(file_name, expected_result)
+
+
+def assert_json_equal(file_name, expected_rows):
+    with open(file_name, "r") as file:
+        data = json.load(file)
+        assert (
+            data == expected_rows
+        ), f"Data in JSON file didn't match expected output: {data} != {expected_rows}"
+
+    os.remove(file_name)

test/detection/test_polygonzone.py

@@ -0,0 +1,94 @@
+from contextlib import ExitStack as DoesNotRaise
+from test.test_utils import mock_detections
+
+import numpy as np
+import pytest
+
+import supervision as sv
+
+DETECTION_BOXES = np.array(
+    [
+        [35.0, 35.0, 65.0, 65.0],
+        [60.0, 60.0, 90.0, 90.0],
+        [85.0, 85.0, 115.0, 115.0],
+        [110.0, 110.0, 140.0, 140.0],
+        [135.0, 135.0, 165.0, 165.0],
+        [160.0, 160.0, 190.0, 190.0],
+        [185.0, 185.0, 215.0, 215.0],
+        [210.0, 210.0, 240.0, 240.0],
+        [235.0, 235.0, 265.0, 265.0],
+    ],
+    dtype=np.float32,
+)
+
+DETECTIONS = mock_detections(
+    xyxy=DETECTION_BOXES, class_id=np.array([0, 0, 0, 0, 0, 0, 0, 0, 0])
+)
+
+POLYGON = np.array([[100, 100], [200, 100], [200, 200], [100, 200]])
+FRAME_RESOLUTION = (300, 300)
+
+
+@pytest.mark.parametrize(
+    "detections, polygon_zone, expected_results, exception",
+    [
+        (
+            DETECTIONS,
+            sv.PolygonZone(
+                POLYGON,
+                FRAME_RESOLUTION,
+                triggering_anchors=(
+                    sv.Position.TOP_LEFT,
+                    sv.Position.TOP_RIGHT,
+                    sv.Position.BOTTOM_LEFT,
+                    sv.Position.BOTTOM_RIGHT,
+                ),
+            ),
+            np.array(
+                [False, False, False, True, True, True, False, False, False], dtype=bool
+            ),
+            DoesNotRaise(),
+        ),  # Test all four corners
+        (
+            DETECTIONS,
+            sv.PolygonZone(
+                POLYGON,
+                FRAME_RESOLUTION,
+            ),
+            np.array(
+                [False, False, True, True, True, True, False, False, False], dtype=bool
+            ),
+            DoesNotRaise(),
+        ),  # Test default behaviour when no anchors are provided
+        (
+            DETECTIONS,
+            sv.PolygonZone(
+                POLYGON,
+                FRAME_RESOLUTION,
+                triggering_anchors=[sv.Position.BOTTOM_CENTER],
+            ),
+            np.array(
+                [False, False, True, True, True, True, False, False, False], dtype=bool
+            ),
+            DoesNotRaise(),
+        ),  # Test default behaviour with deprecated api.
+        (
+            sv.Detections.empty(),
+            sv.PolygonZone(
+                POLYGON,
+                FRAME_RESOLUTION,
+            ),
+            np.array([], dtype=bool),
+            DoesNotRaise(),
+        ),  # Test empty detections
+    ],
+)
+def test_polygon_zone_trigger(
+    detections: sv.Detections,
+    polygon_zone: sv.PolygonZone,
+    expected_results: np.ndarray,
+    exception: Exception,
+) -> None:
+    with exception:
+        in_zone = polygon_zone.trigger(detections)
+        assert np.all(in_zone == expected_results)

test/detection/test_utils.py

@@ -6,13 +6,14 @@ import pytest
 
 from supervision.config import CLASS_NAME_DATA_FIELD
 from supervision.detection.utils import (
+    box_non_max_suppression,
     calculate_masks_centroids,
     clip_boxes,
     filter_polygons_by_area,
     get_data_item,
+    mask_non_max_suppression,
     merge_data,
     move_boxes,
-    non_max_suppression,
     process_roboflow_result,
     scale_boxes,
 )
@@ -113,19 +114,225 @@ TEST_MASK[:, 300:351, 200:251] = True
         ),  # three boxes with different category
     ],
 )
-def test_non_max_suppression(
+def test_box_non_max_suppression(
     predictions: np.ndarray,
     iou_threshold: float,
     expected_result: Optional[np.ndarray],
     exception: Exception,
 ) -> None:
     with exception:
-        result = non_max_suppression(
+        result = box_non_max_suppression(
             predictions=predictions, iou_threshold=iou_threshold
         )
         assert np.array_equal(result, expected_result)
 
 
+@pytest.mark.parametrize(
+    "predictions, masks, iou_threshold, expected_result, exception",
+    [
+        (
+            np.empty((0, 6)),
+            np.empty((0, 5, 5)),
+            0.5,
+            np.array([]),
+            DoesNotRaise(),
+        ),  # empty predictions and masks
+        (
+            np.array([[0, 0, 0, 0, 0.8]]),
+            np.array(
+                [
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, True, False],
+                        [False, True, True, True, False],
+                        [False, True, True, True, False],
+                        [False, False, False, False, False],
+                    ]
+                ]
+            ),
+            0.5,
+            np.array([True]),
+            DoesNotRaise(),
+        ),  # single mask with no category
+        (
+            np.array([[0, 0, 0, 0, 0.8, 0]]),
+            np.array(
+                [
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, True, False],
+                        [False, True, True, True, False],
+                        [False, True, True, True, False],
+                        [False, False, False, False, False],
+                    ]
+                ]
+            ),
+            0.5,
+            np.array([True]),
+            DoesNotRaise(),
+        ),  # single mask with category
+        (
+            np.array([[0, 0, 0, 0, 0.8], [0, 0, 0, 0, 0.9]]),
+            np.array(
+                [
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, False, False],
+                        [False, True, True, False, False],
+                        [False, False, False, False, False],
+                        [False, False, False, False, False],
+                    ],
+                    [
+                        [False, False, False, False, False],
+                        [False, False, False, False, False],
+                        [False, False, False, True, True],
+                        [False, False, False, True, True],
+                        [False, False, False, False, False],
+                    ],
+                ]
+            ),
+            0.5,
+            np.array([True, True]),
+            DoesNotRaise(),
+        ),  # two masks non-overlapping with no category
+        (
+            np.array([[0, 0, 0, 0, 0.8], [0, 0, 0, 0, 0.9]]),
+            np.array(
+                [
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, True, False],
+                        [False, True, True, True, False],
+                        [False, True, True, True, False],
+                        [False, False, False, False, False],
+                    ],
+                    [
+                        [False, False, False, False, False],
+                        [False, False, True, True, True],
+                        [False, False, True, True, True],
+                        [False, False, True, True, True],
+                        [False, False, False, False, False],
+                    ],
+                ]
+            ),
+            0.4,
+            np.array([False, True]),
+            DoesNotRaise(),
+        ),  # two masks partially overlapping with no category
+        (
+            np.array([[0, 0, 0, 0, 0.8, 0], [0, 0, 0, 0, 0.9, 1]]),
+            np.array(
+                [
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, True, False],
+                        [False, True, True, True, False],
+                        [False, True, True, True, False],
+                        [False, False, False, False, False],
+                    ],
+                    [
+                        [False, False, False, False, False],
+                        [False, False, True, True, True],
+                        [False, False, True, True, True],
+                        [False, False, True, True, True],
+                        [False, False, False, False, False],
+                    ],
+                ]
+            ),
+            0.5,
+            np.array([True, True]),
+            DoesNotRaise(),
+        ),  # two masks partially overlapping with different category
+        (
+            np.array(
+                [
+                    [0, 0, 0, 0, 0.8],
+                    [0, 0, 0, 0, 0.85],
+                    [0, 0, 0, 0, 0.9],
+                ]
+            ),
+            np.array(
+                [
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, False, False],
+                        [False, True, True, False, False],
+                        [False, False, False, False, False],
+                        [False, False, False, False, False],
+                    ],
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, False, False],
+                        [False, True, True, False, False],
+                        [False, False, False, False, False],
+                        [False, False, False, False, False],
+                    ],
+                    [
+                        [False, False, False, False, False],
+                        [False, False, False, True, True],
+                        [False, False, False, True, True],
+                        [False, False, False, False, False],
+                        [False, False, False, False, False],
+                    ],
+                ]
+            ),
+            0.5,
+            np.array([False, True, True]),
+            DoesNotRaise(),
+        ),  # three masks with no category
+        (
+            np.array(
+                [
+                    [0, 0, 0, 0, 0.8, 0],
+                    [0, 0, 0, 0, 0.85, 1],
+                    [0, 0, 0, 0, 0.9, 2],
+                ]
+            ),
+            np.array(
+                [
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, False, False],
+                        [False, True, True, False, False],
+                        [False, False, False, False, False],
+                        [False, False, False, False, False],
+                    ],
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, False, False],
+                        [False, True, True, False, False],
+                        [False, True, True, False, False],
+                        [False, False, False, False, False],
+                    ],
+                    [
+                        [False, False, False, False, False],
+                        [False, True, True, False, False],
+                        [False, True, True, False, False],
+                        [False, False, False, False, False],
+                        [False, False, False, False, False],
+                    ],
+                ]
+            ),
+            0.5,
+            np.array([True, True, True]),
+            DoesNotRaise(),
+        ),  # three masks with different category
+    ],
+)
+def test_mask_non_max_suppression(
+    predictions: np.ndarray,
+    masks: np.ndarray,
+    iou_threshold: float,
+    expected_result: Optional[np.ndarray],
+    exception: Exception,
+) -> None:
+    with exception:
+        result = mask_non_max_suppression(
+            predictions=predictions, masks=masks, iou_threshold=iou_threshold
+        )
+        assert np.array_equal(result, expected_result)
+
+
 @pytest.mark.parametrize(
     "xyxy, resolution_wh, expected_result",
     [
@@ -838,7 +1045,7 @@ def test_merge_data(
                 "test_1": [1],
             },
             DoesNotRaise(),
-        ),  # single data dict with a single field name and list values
+        ),  # data dict with a single list field and integer index
         (
             {
                 "test_1": np.array([1, 2, 3]),
@@ -848,7 +1055,7 @@ def test_merge_data(
                 "test_1": np.array([1]),
             },
             DoesNotRaise(),
-        ),  # single data dict with a single field name and np.array values as 1D arrays
+        ),  # data dict with a single np.array field and integer index
         (
             {
                 "test_1": [1, 2, 3],
@@ -858,7 +1065,7 @@ def test_merge_data(
                 "test_1": [1, 2],
             },
             DoesNotRaise(),
-        ),  # single data dict with a single field name and list values
+        ),  # data dict with a single list field and slice index
         (
             {
                 "test_1": np.array([1, 2, 3]),
@@ -868,7 +1075,7 @@ def test_merge_data(
                 "test_1": np.array([1, 2]),
             },
             DoesNotRaise(),
-        ),  # single data dict with a single field name and np.array values as 1D arrays
+        ),  # data dict with a single np.array field and slice index
         (
             {
                 "test_1": [1, 2, 3],
@@ -878,7 +1085,7 @@ def test_merge_data(
                 "test_1": [3],
             },
             DoesNotRaise(),
-        ),  # single data dict with a single field name and list values
+        ),  # data dict with a single list field and negative integer index
         (
             {
                 "test_1": np.array([1, 2, 3]),
@@ -888,7 +1095,7 @@ def test_merge_data(
                 "test_1": np.array([3]),
             },
             DoesNotRaise(),
-        ),  # single data dict with a single field name and np.array values as 1D arrays
+        ),  # data dict with a single np.array field and negative integer index
         (
             {
                 "test_1": [1, 2, 3],
@@ -898,7 +1105,7 @@ def test_merge_data(
                 "test_1": [1, 3],
             },
             DoesNotRaise(),
-        ),  # single data dict with a single field name and list values
+        ),  # data dict with a single list field and integer list index
         (
             {
                 "test_1": np.array([1, 2, 3]),
@@ -908,7 +1115,7 @@ def test_merge_data(
                 "test_1": np.array([1, 3]),
             },
             DoesNotRaise(),
-        ),  # single data dict with a single field name and np.array values as 1D arrays
+        ),  # data dict with a single np.array field and integer list index
         (
             {
                 "test_1": [1, 2, 3],
@@ -918,7 +1125,7 @@ def test_merge_data(
                 "test_1": [1, 3],
             },
             DoesNotRaise(),
-        ),  # single data dict with a single field name and list values
+        ),  # data dict with a single list field and integer np.array index
         (
             {
                 "test_1": np.array([1, 2, 3]),
@@ -928,7 +1135,55 @@ def test_merge_data(
                 "test_1": np.array([1, 3]),
             },
             DoesNotRaise(),
-        ),
+        ),  # data dict with a single np.array field and integer np.array index
+        (
+            {
+                "test_1": np.array([1, 2, 3]),
+            },
+            np.array([True, True, True]),
+            {
+                "test_1": np.array([1, 2, 3]),
+            },
+            DoesNotRaise(),
+        ),  # data dict with a single np.array field and all-true bool np.array index
+        (
+            {
+                "test_1": np.array([1, 2, 3]),
+            },
+            np.array([False, False, False]),
+            {
+                "test_1": np.array([]),
+            },
+            DoesNotRaise(),
+        ),  # data dict with a single np.array field and all-false bool np.array index
+        (
+            {
+                "test_1": np.array([1, 2, 3]),
+            },
+            np.array([False, True, False]),
+            {
+                "test_1": np.array([2]),
+            },
+            DoesNotRaise(),
+        ),  # data dict with a single np.array field and mixed bool np.array index
+        (
+            {"test_1": np.array([1, 2, 3]), "test_2": ["a", "b", "c"]},
+            0,
+            {"test_1": np.array([1]), "test_2": ["a"]},
+            DoesNotRaise(),
+        ),  # data dict with two fields and integer index
+        (
+            {"test_1": np.array([1, 2, 3]), "test_2": ["a", "b", "c"]},
+            -1,
+            {"test_1": np.array([3]), "test_2": ["c"]},
+            DoesNotRaise(),
+        ),  # data dict with two fields and negative integer index
+        (
+            {"test_1": np.array([1, 2, 3]), "test_2": ["a", "b", "c"]},
+            np.array([False, True, False]),
+            {"test_1": np.array([2]), "test_2": ["b"]},
+            DoesNotRaise(),
+        ),  # data dict with two fields and mixed bool np.array index
     ],
 )
 def test_get_data_item(