| AnthroTAP: Learning Point Tracking with Real-World Motion | | 0.714 | 0.924 | 0.835 | Anthro-TAPNext (extra data) | 2025-07-08 |
| TAPNext: Tracking Any Point (TAP) as Next Token Prediction | ✓ Link | 0.689 | 0.916 | 0.824 | BootsTAPNext-B (extra data) | 2025-04-08 |
| Dense Optical Tracking: Connecting the Dots | ✓ Link | 0.685 | 0.917 | 0.813 | DOT (BootsTAPIR + RAFT) (extra data) | 2023-12-01 |
| Local All-Pair Correspondence for Point Tracking | ✓ Link | 0.678 | 0.899 | 0.796 | LocoTrack-B (extra data) | 2024-07-22 |
| BootsTAP: Bootstrapped Training for Tracking-Any-Point | ✓ Link | 0.674 | 0.913 | 0.79 | BootsTAPIR+fix+snap+data (extra data) | 2024-02-01 |
| Online Dense Point Tracking with Streaming Memory | ✓ Link | 0.669 | 0.902 | 0.794 | SPOT | 2025-03-09 |
| TAPTRv2: Attention-based Position Update Improves Tracking Any Point | | 0.664 | 0.913 | 0.788 | TAPTRv2 | 2024-07-23 |
| TAPTR: Tracking Any Point with Transformers as Detection | ✓ Link | 0.663 | 0.91 | 0.792 | TAPTR | 2024-03-19 |
| ProTracker: Probabilistic Integration for Robust and Accurate Point Tracking | ✓ Link | 0.653 | 0.887 | 0.808 | ProTracker | 2025-01-06 |
| DINO-Tracker: Taming DINO for Self-Supervised Point Tracking in a Single Video | ✓ Link | 0.623 | 0.875 | 0.782 | DINO-Tracker | 2024-03-21 |
| TAPIR: Tracking Any Point with per-frame Initialization and temporal Refinement | ✓ Link | 0.613 | 0.888 | 0.736 | TAPIR (Panning MOVi-E) (extra data) | 2023-06-14 |
| MFT: Long-Term Tracking of Every Pixel | ✓ Link | 0.561 | 0.869 | 0.708 | MFT | 2023-05-22 |
| Tracking Everything Everywhere All at Once | ✓ Link | 0.517 | 0.853 | 0.675 | OmniMotion (RAFT) | 2023-06-08 |
| TAP-Vid: A Benchmark for Tracking Any Point in a Video | ✓ Link | 0.384 | 0.823 | 0.531 | TAP-Net | 2022-11-07 |