Research map / 3D vision / Human pose estimation
Mocap-2-to-3: Multi-view Lifting for Monocular Motion Recovery with 2D Pretraining
Zhumei Wang, Zechen Hu, Ruoxi Guo, Huaijin Pi, Ziyong Feng, Liang Zhang, Mingtao Pei, Siyuan Huang
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 |
|---|---|---|---|
| decouple local pose and global movement representation (ablation) | Mocap-2-to-3 · RICH | PA-MPJPE (mm): 65.1 -> 30.5 | Improved |
| add ground-plane pointmaps (ablation) | Mocap-2-to-3 · RICH | PA-MPJPE (mm) at 3.5k epochs: 45.8 -> 30.5 | Improved |
| add ground-plane pointmaps (ablation) | Mocap-2-to-3 · RICH | PA-MPJPE (mm) at 8k epochs: 33.4 (8k, w/o) vs 30.5 (3.5k, w/) | No clear effect |
| add in-domain 2D RICH sequences (175) to 2D pretraining (ablation) | Mocap-2-to-3 · RICH | PA-MPJPE (mm): 30.5 -> 26.2 | Improved |
| add in-domain 2D RICH sequences to pretraining (ablation) | Mocap-2-to-3 · RICH | Abs-MPJPE (mm): 157.9 -> 156.8 | No clear effect |
| use Mocap-2-to-3 multi-view lifting instead of GVHMR+SMPLify | lifting from GT 2D keypoints · RICH | Abs-MPJPE (mm): 430.4 -> 156.8 | Improved |
| use Mocap-2-to-3 instead of MVLift | 2D-to-3D lifting · AIST++ | PA-MPJPE (mm): 79.2 -> 60.1 | Improved |
| use Mocap-2-to-3 instead of GVHMR | video HMR · RICH | foot sliding: 2.5 -> 3.5 | Hurt |
Closest papers by meaning
- MonoCap: Monocular Human Motion Capture using a CNN Coupled with a Geometric Prior (TPAMI 2018 · similarity 0.91)
- Lifting Motion to the 3D World via 2D Diffusion (CVPR 2025 · similarity 0.90)
- VideoMDM: Towards 3D Human Motion Generation From 2D Supervision (NeurIPS 2026 · similarity 0.88)
- PandaPose: 3D Human Pose Lifting from a Single Image via Propagating 2D Pose Prior to 3D Anchor Space (NeurIPS 2025 · similarity 0.88)
- HumanMM: Global Human Motion Recovery from Multi-shot Videos (CVPR 2025 · similarity 0.88)
- Lifting by Image – Leveraging Image Cues for Accurate 3D Human Pose Estimation (AAAI 2024 · similarity 0.88)
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.
