yolov7-object-tracking

Why we included this project

This is a compact YOLOv7 + SORT tracking reference that handles detection and association for you, so you can add persistent object tracking to a video pipeline without wiring up the tracking logic yourself. It works with video files, webcams, external cameras, and IP streams, and it can filter detections by class, assign each track its own color, and dump track IDs, centroids, and bounding-box coordinates to text files. The code is small enough to read through in one sitting, which makes it a decent starting point for prototyping things like traffic monitoring or people counting. It also accepts YOLOv8 weights through the same detect script, so you can compare detector versions without changing tools. If you want a self-contained example rather than a full framework, this is easy to adapt.