All Tools
S
OtherFreeOpen Source
SUPERVISION
We write your reusable computer vision tools
MIT
ABOUT
Traditional computer vision workflows require custom code for visualizing detections, tracking objects, and managing datasets — forcing developers to reimplement boilerplate for every project. Supervision provides a unified, model-agnostic toolkit with ready-made annotators (boxes, masks, labels, keypoints), ByteTrack integration, zone-based counting, dataset converters (COCO, YOLO, Pascal VOC), and speed estimation that works across popular CV frameworks. This cuts development time from days to minutes for common computer vision tasks.
INSTALL
pip install supervisionINTEGRATION GUIDE
1. Object detection: draw bounding boxes, masks, labels, and keypoints on images and video streams with customizable annotators
2. Multi-object tracking with ByteTrack for consistent object IDs across video frames
3. Zone-based counting: track objects entering and exiting defined polygonal zones for traffic or retail analytics
4. Speed estimation using perspective transformation and tracking data for vehicle monitoring
5. Dataset management: load, split, merge, and save datasets in COCO, YOLO, and Pascal VOC formats
6. Video processing with detection, tracking, and annotation applied frame-by-frame to live or recorded streams
TAGS
computer-visionobject-detectionannotationtrackingpythonopen-sourcevideo-processingyolo