ByteTrack is a multi-object tracking method that improves tracking performance by associating every detection box, not just high-score ones. The method, named BYTE, uses both high and low score detection boxes to recover true objects and filter out background detections. It is applied to nine state-of-the-art trackers, achieving consistent improvements in metrics such as MOTA, IDF1, and HOTA. ByteTrack achieves 80.3 MOTA, 77.3 IDF1, and 63.1 HOTA on the MOT17 test set with 30 FPS speed on a single V100 GPU, and also performs well on MOT20, HiEve, and BDD100K benchmarks. The method is simple, effective, and robust to occlusion, and can be easily integrated into existing trackers. The source code and pre-trained models are available for use.ByteTrack is a multi-object tracking method that improves tracking performance by associating every detection box, not just high-score ones. The method, named BYTE, uses both high and low score detection boxes to recover true objects and filter out background detections. It is applied to nine state-of-the-art trackers, achieving consistent improvements in metrics such as MOTA, IDF1, and HOTA. ByteTrack achieves 80.3 MOTA, 77.3 IDF1, and 63.1 HOTA on the MOT17 test set with 30 FPS speed on a single V100 GPU, and also performs well on MOT20, HiEve, and BDD100K benchmarks. The method is simple, effective, and robust to occlusion, and can be easily integrated into existing trackers. The source code and pre-trained models are available for use.