Multiple Object Tracking On Bdd100K Val
Metrics
TETA
mIDF1
mMOTA
Results
Performance results of various models on this benchmark
| Paper Title | ||||
|---|---|---|---|---|
| ByteTrack | 55.7 | 54.8 | 45.5 | ByteTrack: Multi-Object Tracking by Associating Every Detection Box |
| QDTrack | 51.3 | 54.3 | 42.1 | QDTrack: Quasi-Dense Similarity Learning for Appearance-Only Multiple Object Tracking |
| TETer | 50.8 | 53.3 | 39.1 | Tracking Every Thing in the Wild |
| QDTrack | 47.8 | 50.8 | 36.6 | Quasi-Dense Similarity Learning for Multiple Object Tracking |
| Unicorn | - | 54.0 | 41.2 | Towards Grand Unification of Object Tracking |
| ContrasTR | - | 52.9 | 41.7 | Contrastive Learning for Multi-Object Tracking with Transformers |
| UNINEXT-H | - | 56.7 | 44.2 | Universal Instance Perception as Object Discovery and Retrieval |
| MOTRv2 | - | 56.5 | 43.6 | MOTRv2: Bootstrapping End-to-End Multi-Object Tracking by Pretrained Object Detectors |
| Adaptive-searching -windows Tracker | - | 56.5 | 34.4 | Tackling multiple object tracking with complicated motions—Re-designing the integration of motion and appearance |
0 of 9 row(s) selected.