Research map / 3D vision / Human pose estimation
Tracking People with 3D Representations
Jathushan Rajasegaran, Georgios Pavlakos, Angjoo Kanazawa, Jitendra Malik
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Use 3D texture-map appearance instead of 2D Re-ID (OSNet) appearance | single appearance cue · PoseTrack | IDs: 892 -> 783 | Improved |
| Use 3D texture-map appearance instead of Tracktor Re-ID appearance | single appearance cue · AVA | IDF1: 64.6 -> 64.2 | No clear effect |
| Use 3D pose and location instead of 2D pose and location | single pose/location cue · PoseTrack | IDs: 1872 -> 1563 | Improved |
| Use 3D pose and location instead of 2D | single pose/location cue · AVA | IDF1: 56.6 -> 64.2 | Improved |
| Use 3D pose and location instead of 2D | single pose/location cue · MuPoTs | IDF1: 80.8 -> 80.1 | Hurt |
| Add 3D appearance cue (remove it in ablation) (ablation) | Full system · PoseTrack | IDs: 1783 -> 655 | Improved |
| Add 3D appearance cue (remove it in ablation) (ablation) | Full system · AVA | IDF1: 48.7 -> 61.3 | Improved |
| Add 3D location cue (remove it in ablation) (ablation) | Full system · PoseTrack | IDs: 1315 -> 655 | Improved |
| Add 3D pose cue (remove it in ablation) (ablation) | Full system · PoseTrack | IDs: 711 -> 655 | Improved |
| Add 3D pose cue (remove it in ablation) (ablation) | Full system · AVA | IDs: 355 -> 240 | Improved |
| Add 3D location cue (remove it in ablation) (ablation) | Full system · AVA | IDs: 273 -> 240 | Improved |
| Add spatio-temporal transformer aggregation (Ours-v1 to Ours-v2) (ablation) | Full system · PoseTrack | IDF1: 64.5 -> 73.4 | Improved |
Closest papers by meaning
- Tracking People by Predicting 3D Appearance, Location and Pose (CVPR 2022 · similarity 0.91)
- Humans in 4D: Reconstructing and Tracking Humans with Transformers (ICCV 2023 · similarity 0.89)
- Self-supervised Multi-view Person Association and Its Applications (TPAMI 2020 · similarity 0.88)
- Multi-THuMBS: Multi-person Tracking of 3D Human Meshes Beyond Video Shots (ECCV 2026 · similarity 0.88)
- DETRAM: End-to-end DEtection, Tracking and Recovery of HumAn Meshes (ECCV 2026 · similarity 0.88)
- CoMotion: Concurrent Multi-person 3D Motion (ICLR 2025 · similarity 0.87)
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.
