Revise the contents of the 'How to Track Objects' documentation covering the use of supervision for object tracking in video analysis. The revision offers a more detailed guide, including running inference, tracking objects, annotating videos with unique tracking IDs, and annotating videos with traces. The new guide provides a structured walkthrough and tutorial on using Supervision's functionalities, aimed at helping users to annotate videos more effectively and understand the movement patterns and interactions of objects in the videos.
Expanded the documentation to include instructions on how to install and use the Supervision assets utility. Instructions include steps for performing the installation using pip, allowing users to easily download video files for their demos.
The file asset_list.py has been renamed to list.py for better readability and simplicity. In the VideoAssets enumeration, additional information has been included. This includes specific details for each member of the enum, notably the filename of the related video and its URL.
The custom 404.html was removed as it was no longer necessary with the updates to the Mkdocs.yml file. Upgraded the project version from 0.16.0rc2 to 0.16.0rc3 in the pyproject.toml file. Various updates were added to the changelog.md to reflect improvements and bug fixes in the new project release. This includes new annotations, utility improvements, and bug-fixes to improve project functionality.
Expanded the drawing functionality by adding new utility methods 'draw_image', 'draw_line' and 'draw_rectangle' and imported them in supervision/__init__.py. Updated the documentation of draw/utils, to include 'draw_image'. Improved type hinting in draw/color.py and draw/utils.py. Also, some redundant comments and unnecessary immediate memory freeing line are removed from draw/utils.py to enhance code readability.
The hyperlink to images in the example blocks in `supervision/annotators/core.py`, and `docs/annotators.md` was modified to reflect recent updates to the annotated heat map image.
Code examples in `supervision/annotators/core.py` and `docs/annotators.md` have been updated for clarity and consistency. The changes include renaming variables more appropriately and elaborating on the video processing steps which now includes tracking and annotating frames using a YOLO model. This revision aims to make examples more understandable and consistent for users following the documentation.
This commit clarifies the process of object detection and annotation using Supervision and the Ultralytics YOLOv8 model. The changes also include updated instructions on how to load model predictions into Supervision and using them to annotate images, with links to the methods used in the process and the available options. An image showing the resulting annotated image has also been added for better understanding. The changes are shaped to make the documentation more informative, concise, and easy to follow.
Added detailed comments and examples to the Color and ColorPalette classes in supervision/draw/color.py, in addition to the DotAnnotator class in supervision/annotators/core.py. The changes clarify the usage and function of these classes. Also appended to the documentation by creating a new section for Color in docs/draw/color.md and including Color and DotAnnotator in docs/annotators.md. Incorporated the new documentation into the navigation structure in mkdocs.yml for user accessibility. The improvements enhance overall code-readability and understandability.
Descriptions for various attributes and functionalities have been added into LineZone class to aid developers in understanding their behaviour. Also, the trigger function was modified to return two separate boolean arrays - 'crossed_in' and 'crossed_out' to denote objects crossing the line in specific directions. Moreover, Line Zone documentation is added to the side navigation in mkdocs.yml.
Moved advanced filtering documentation to `how_to` directory from `quickstart` to make way for more comprehensive starter guide. Reorganized MkDocs configuration accordingly. Formatted some code in `classification/core.py` and `dataset/core.py` for better readability. Added placeholder files in `how_to` for future guides on object detection, video processing, object tracking, model evaluation, and detections filtering.
Piotr Skalski
pre-commit-ci[bot]
James Gallagher
Onuralp SEZER
Rajarshi Misra