Pose Tracking
Pose Tracking is the task of estimating multi-person human poses in videos and assigning unique instance IDs for each keypoint across frames. Accurate estimation of human keypoint-trajectories is useful for human action recognition, human interaction understanding, motion capture and animation.
Source: LightTrack: A Generic Framework for Online Top-Down Human Pose Tracking
Papers
Showing 1–25 of 191 papers
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | DetTrack | MOTA | 64.09 | — | Unverified |
| 2 | KeyTrack | MOTA | 61.15 | — | Unverified |
| 3 | LightTrack | MOTA | 58.01 | — | Unverified |
| 4 | HRNet-W48 COCO | MOTA | 57.93 | — | Unverified |
| 5 | MSRA (FlowTrack) | MOTA | 57.81 | — | Unverified |
| 6 | TML++ (MIPAL) | MOTA | 54.46 | — | Unverified |
| 7 | STAF | MOTA | 53.81 | — | Unverified |
| 8 | ProTracker | MOTA | 51.82 | — | Unverified |
| 9 | PoseFlow | MOTA | 50.98 | — | Unverified |
| 10 | PoseTrack | MOTA | 48.37 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | DetTrack | MOTA | 64.3 | — | Unverified |
| 2 | UniTrack | MOTA | 63.5 | — | Unverified |
| 3 | 4DHumans + ViTDet | MOTA | 61.9 | — | Unverified |
| 4 | MSRA | MOTA | 61.37 | — | Unverified |
| 5 | TML++ (MIPAL) | MOTA | 54.86 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | PoseTrack | MOTA | 28.2 | — | Unverified |