Research map / Detection and segmentation / Tracking
PoseTrack21: A Dataset for Person Search, Multi-Object Tracking and Multi-Person Pose Tracking
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add ReID track-history matching (ablation) | CorrTrack pose tracking · PoseTrack21 pose tracking | MOTA: 63.0 -> 63.8 | Helped |
| Add ReID track-history matching (ablation) | CorrTrack pose tracking · PoseTrack21 pose tracking | HOTA: 51.13 -> 52.71 | Helped |
| Use offline track merging (ablation) | CorrTrack pose tracking · PoseTrack21 pose tracking | MOTA: 63.0 -> 63.9 | Helped |
| Use offline track merging (ablation) | CorrTrack pose tracking · PoseTrack21 pose tracking | HOTA: 51.13 -> 52.42 | Helped |
| Replace the motion model with keypoint correspondences (ablation) | Tracktor++ pose tracking · PoseTrack21 pose tracking | HOTA: 52.21 -> 48.90 | Hurt |
| Add pose estimation to Tracktor++ (ablation) | Tracktor++ · PoseTrack21 pose tracking | MOTA: Tracktor++ w. poses: 63.3 | Helped |
| Use CorrTrack with ReID outputs as bounding-box tracks (ablation) | Multi-object tracking on PoseTrack21MOT · PoseTrack21MOT | HOTA: 56.95; 58.29 | Mixed |
| Use SeqNet rather than the other evaluated person-search baselines | Person search · PoseTrack21 person search | mAP: 57.58 -> 65.12 | Helped |
| Use the NAE+ pixel-wise extension (ablation) | NAE person search · PoseTrack21 person search | mAP: 55.83 -> 54.48 | Hurt |
Closest papers by meaning
- PoseTrack: Joint Multi-Person Pose Estimation and Tracking (CVPR 2017 · similarity 0.95)
- PoseTrack: A Benchmark for Human Pose Estimation and Tracking (CVPR 2018 · similarity 0.94)
- ArtTrack: Articulated Multi-Person Tracking in the Wild (CVPR 2017 · similarity 0.93)
- Combining Detection and Tracking for Human Pose Estimation in Videos (CVPR 2020 · similarity 0.91)
- Detect-and-Track: Efficient Pose Estimation in Videos (CVPR 2018 · similarity 0.91)
- CoMotion: Concurrent Multi-person 3D Motion (ICLR 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.