Research map / 3D vision / Human pose estimation
PressTrack-HMR: Pressure-Based Top-Down Multi-Person Global Human Mesh Recovery
Jiayue Yuan, Fangting Xie, Guangwen Ouyang, Changhai Ma, Ziyu Wu, Heyu Ding, Quan Wan, Yi Ke, Yuchen Wu, Xiaohui Cai
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 UoE-based association without motion prediction instead of ByteTrack/BoT-SORT | pressure footprint tracking · MIP | MOTA: 82.5% -> 93.6% (BoT-SORT vs ours) | Improved |
| use UoE-based tracker instead of BoT-SORT | pressure footprint tracking · MIP | ID switches: 10494 -> 437 | Improved |
| increase temporal input sequence length from 1 to 16 (ablation) | HMR module · MIP (ground-truth detections) | MPJPE (mm): 94.2 -> 81.8 | Improved |
| increase temporal input sequence length from 1 to 16 (ablation) | PressTrack-HMR end-to-end · MIP (tracked detections) | MPJPE (mm): 96.2 -> 89.2 | Improved |
| increase sequence length from 16 to 32 (ablation) | HMR module · MIP (ground-truth detections) | MPJPE (mm): 81.8 -> 82.1 | Hurt |
| increase sequence length from 16 to 32 (ablation) | PressTrack-HMR end-to-end · MIP (tracked detections) | MPJPE (mm): 89.2 -> 89.7 | Hurt |
| use tracked detections instead of ground-truth detections as HMR input (ablation) | PressTrack-HMR · MIP unseen sequence | WA-MPJPE100 (mm): 90.9 -> 112.6 | Hurt |
| evaluate on unseen subjects instead of unseen sequences (ablation) | PressTrack-HMR end-to-end · MIP | MPJPE (mm): 89.2 -> 96.8 | Hurt |
Closest papers by meaning
- MotionPRO: Exploring the Role of Pressure in Human MoCap and Beyond (CVPR 2025 · similarity 0.87)
- PI-HMR: Towards Robust In-bed Temporal Human Shape Reconstruction with Contact Pressure Sensing (CVPR 2025 · similarity 0.86)
- Harmony4D: A Video Dataset for In-The-Wild Close Human Interactions (NeurIPS 2024 · similarity 0.85)
- Optimizing Human Pose Estimation Through Focused Human and Joint Regions (AAAI 2025 · similarity 0.85)
- Motions as Queries: One-Stage Multi-Person Holistic Human Motion Capture (CVPR 2025 · similarity 0.84)
- SAT-HMR: Real-Time Multi-Person 3D Mesh Estimation via Scale-Adaptive Tokens (CVPR 2025 · similarity 0.84)
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.
