Point Tracking on DAVIS
7.7AJCRW-C
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CRW-CSupervision Type=Self-supervised2025.12 | 7.7 | 13.5 | 72.9 | |
| DIFT-CSupervision Type=Self-supervised2025.12 | 18.1 | 33 | 68.8 | |
| CRW-DSupervision Type=Self-supervised2025.12 | 23.6 | 38 | 77.2 | |
| Flow-Walk-DSupervision Type=Self-supervised2025.12 | 24.4 | 40.9 | 76.5 | |
| DIFT-DSupervision Type=Self-supervised2025.12 | 29.7 | 48.2 | 77.2 | |
| GMRW-DSupervision Type=Self-supervised2025.12 | 30.3 | 49.4 | 77.3 | |
| Kubric-VFS-LikeSupervision Type=Supervised2025.12 | 33.1 | 48.5 | 79.4 | |
| ARFlow-CSupervision Type=Self-supervised2025.12 | 35 | 51.8 | 79.7 | |
| Flow-Walk-CSupervision Type=Self-supervised2025.12 | 35.2 | 51.4 | 80.6 | |
| COTRSupervision Type=Supervised2025.12 | 35.4 | 51.3 | 80.2 | |
| TAP-NetSupervision Type=Supervised2025.12 | 38.4 | 53.1 | 82.3 | |
| Video-GMAEModel Variant=zeroshot, Stride=52025.12 | 41.3 | 55.7 | 85.2 | |
| GMRW-CSupervision Type=Self-supervised2025.12 | 41.8 | 60.9 | 78.3 | |
| PIPsSupervision Type=Supervised2025.12 | 42 | 59.4 | 82.1 | |
| Video-GMAEModel Variant=base, Evaluation Protocol=frozen, Stride=52025.12 | 45.2 | 55.2 | 91.4 | |
| Video-GMAEModel Variant=large, Evaluation Protocol=frozen, Stride=52025.12 | 46.7 | 55.8 | 91.8 | |
| Video-GMAEModel Variant=base, Evaluation Protocol=finetune, Stride=52025.12 | 55.7 | 66.1 | 92.1 | |
| TAPIRTrain data=Kub, Track Type=Sparse2026.02 | 56.2 | 70 | 86.5 | |
| TAPIRSupervision Type=Supervised2025.12 | 56.2 | 70 | 86.5 | |
| Video-GMAEModel Variant=large, Evaluation Protocol=finetune, Stride=52025.12 | 57.9 | 67.7 | 93.5 | |
| DOTTrain data=Kub, Track Type=Dense2026.02 | 60.1 | 74.5 | 89 | |
| DELTATrain data=Kub, Track Type=Dense2026.02 | 60.8 | 75.2 | 87.6 | |
| BootsTAPIRTrain data=Kub+15M, Track Type=Sparse2026.02 | 61.4 | 73.6 | 88.7 | |
| BootsTAPIRSupervision Type=Supervised2025.12 | 61.4 | 73.6 | 88.7 | |
| CoTrackerTrain data=Kub, Track Type=Sparse2026.02 | 61.8 | 76.1 | 88.3 | |
| CoTrackerSupervision Type=Supervised2025.12 | 61.8 | 76.1 | 88.3 | |
| AllTrackerTrain data=Kub, Track Type=Dense2026.02 | 61.9 | 75.4 | 87.8 | |
| LocoTrackTrain data=Kub, Track Type=Sparse2026.02 | 62.9 | 75.3 | 87.2 | |
| LocoTrackSupervision Type=Supervised2025.12 | 62.9 | 75.3 | 87.2 | |
| TAPTRTrain data=Kub, Track Type=Sparse2026.02 | 63 | 76.1 | 91.1 | |
| AllTrackerTrain data=Kub+Mix, Track Type=Dense2026.02 | 63.3 | 76.3 | 90.1 | |
| CoTracker3Train data=Kub+15k, Track Type=Sparse, Inference Mode=online2026.02 | 63.8 | 76.3 | 90.2 | |
| CoTracker3Supervision Type=Supervised2025.12 | 63.8 | 76.3 | 90.2 | |
| CoTracker3Train data=Kub+15k, Track Type=Sparse, Inference Mode=offline2026.02 | 64.4 | 76.9 | 91.2 | |
| CoWTrackerTrain data=Kub, Track Type=Dense2026.02 | 65.5 | 78 | 92.1 | |
| MooGParam (M)=35, frozen representations=true, readout head=from MooG, num_frames=82024.12 | 68.7 | — | — | |
| VideoMAEParam (M)=330, frozen representations=true, readout head=from MooG, num_frames=82024.12 | 70.3 | — | — | |
| TRecViTParam (M)=128, frozen representations=true, readout head=from MooG, num_frames=82024.12 | 70.6 | — | — | |
| PIPs++Train data=PO, Track Type=Sparse2026.02 | — | 73.7 | — | |
| PIPs++Supervision Type=Supervised2025.12 | — | 73.7 | — |