Tracking by Detection and Query: An Efficient End-to-End Framework for Multi-Object Tracking

TBDQ-Net advances multi-object tracking by merging tracking-by-detection and tracking-by-query methods. It uses a frozen detector and a lightweight associator to efficiently link detection queries (new objects) with track queries (existing objects). This results in better accuracy than existing TBD methods while significantly reducing model size.

Open the full episode

More from AI Radio