Multi-object tracking with YOLOv8, but it memorizes previously tracked objects and recognizes them. Each detected object's embedding is computed and then stored as a hash. It uses Locality Sensitive Hashing (LSH) for approximate nearest neighbor search of hashes. Similar objects are stored in the same "buckets".
BigRedDoge/MemTrack
Multi-object tracking with YOLOv8, but it remembers and recognizes previously tracked objects
Python