Published methods evaluated on the DanceTrack test set. Results are from the CodaBench evaluation server.
For a sortable table (click any column header to rank), see LEADERBOARD.csv. GitHub renders CSV files with built-in column sorting.
Please add your result here according to the publication date order (the newer the later).
| Method | HOTA | DetA | AssA | MOTA | IDF1 | CodaBench ID | Paper | Venue | Date |
|---|---|---|---|---|---|---|---|---|---|
| OC-SORT | 55.7 | 81.7 | 38.1 | 92.0 | 54.6 | 628670 | arXiv | CVPR 2023 | 2023-06 |
- HOTA (Higher Order Tracking Accuracy): Primary ranking metric. Balances detection and association.
- DetA (Detection Accuracy): Measures detection quality.
- AssA (Association Accuracy): Measures identity association quality.
- MOTA (Multiple Object Tracking Accuracy): Classic metric from CLEAR MOT.
- IDF1 (ID F1 Score): Measures identity preservation.
To add your method to this leaderboard:
- Submit results to CodaBench and make it public.
- Open a pull request (with "[LEADERBOARD]" in the title) adding a row to both
LEADERBOARD.mdandLEADERBOARD.csvwith your method name, CodaBench submission ID, performance metrics, paper link, venue, and publication date.