History

Paul f942d89881 Added inference examples throughout the docs.		2023-12-22 11:09:39 -07:00
..
README.md	Added inference examples throughout the docs.	2023-12-22 11:09:39 -07:00
inference_example.py	Added inference examples throughout the docs.	2023-12-22 11:09:39 -07:00
requirements.txt	fix(pre_commit): 🎨 auto format pre-commit hooks	2023-12-15 16:42:53 +00:00
ultralytics_example.py	Added examples using `inference` to each example directory and updated requirements and READMEs accordingly.	2023-12-15 09:38:05 -07:00

README.md

tracking

👋 hello

This script provides functionality for processing videos using YOLOv8 for object detection and Supervision for tracking and annotation.

💻 install

clone repository and navigate to example directory

git clone https://github.com/roboflow/supervision.git
cd supervision/examples/tracking

setup python environment and activate it [optional]
```
python3 -m venv venv
source venv/bin/activate
```
install required dependencies
```
pip install -r requirements.txt
```

🛠️ script arguments

inference args

--roboflow_api_key: Your Roboflow API key
--model_id (optional): Specifies the Roboflow model id (dataset/version) to use for inference. See COCO models on Roboflow Universe. Default is yolov8x-1280.
--source_video_path: Required. The path to the source video file to be processed. This is the video on which object detection and annotation will be performed.
--target_video_path: Required. The path where the processed video, with annotations added, will be saved. This is your output video file.
--confidence_threshold (optional): Sets the confidence level at which the model identifies objects in the video. Default is 0.3. A higher threshold makes the model more selective, while a lower threshold makes it more inclusive in identifying objects.
--iou_threshold (optional): Specifies the IOU (Intersection Over Union) threshold for the model, defaulting to 0.7. This parameter helps in differentiating between distinct objects, especially in crowded scenes.

ultralytics args

--source_weights_path: Required. Specifies the path to the YOLO model's weights file, which is essential for the object detection process. This file contains the data that the model uses to identify objects in the video.
--source_video_path: Required. The path to the source video file to be processed. This is the video on which object detection and annotation will be performed.
--target_video_path: Required. The path where the processed video, with annotations added, will be saved. This is your output video file.
--confidence_threshold (optional): Sets the confidence level at which the model identifies objects in the video. Default is 0.3. A higher threshold makes the model more selective, while a lower threshold makes it more inclusive in identifying objects.
--iou_threshold (optional): Specifies the IOU (Intersection Over Union) threshold for the model, defaulting to 0.7. This parameter helps in differentiating between distinct objects, especially in crowded scenes.

⚙️ run

inference

python inference_example.py \
--roboflow_api_key <ROBOFLOW API KEY> \
--source_video_path input.mp4 \
--target_video_path tracking_result.mp4

ultralytics

python ultralytics_example.py \
--source_weights_path yolov8s.pt \
--source_video_path input.mp4 \
--target_video_path tracking_result.mp4

© license

This demo integrates two main components, each with its own licensing:

YOLOv8: The object detection model used in this demo, YOLOv8, is distributed under the AGPL-3.0 license. You can find more details about this license here.

Supervision: The analytics code that powers the tracking and visualization in this demo is based on the Supervision library, which is licensed under the MIT license. This makes the Supervision part of the code fully open source and freely usable in your projects.