Research map / 3D vision / Human pose estimation
Physics-Based Human Motion Estimation and Synthesis From Videos
Kevin Xie, Tingwu Wang, Umar Iqbal, Yunrong Guo, Sanja Fidler, Florian Shkurti
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 |
|---|---|---|---|
| add physics loss to kinematic optimization (ablation) | pose optimization · Human3.6M (S9, S11) | MPJPE (mm): 73.6 -> 68.1 | Improved |
| add physics loss to kinematic optimization (ablation) | pose optimization · Human3.6M (S9, S11) | global root position error (mm): 148.2 -> 85.1 | Improved |
| add physics loss to kinematic optimization (ablation) | pose optimization · Human3.6M (S9, S11) | e_smooth: 5.42 -> 4.0 | Improved |
| add physics loss to kinematic optimization (ablation) | pose optimization · Human3.6M (S9, S11) | sigma_smooth: 1.06 -> 1.3 | Hurt |
| add physics loss to kinematic optimization (ablation) | pose optimization · Human3.6M (S9, S11) | foot tangential velocity error: 4.65 -> 2.71 | Improved |
| add physics loss to kinematic optimization (ablation) | pose optimization · Human3.6M (S9, S11) | foot global height error: 95.7 -> 18.9 | Improved |
| train DLow on physics-corrected rather than kinematic pose estimates (ablation) | DLow · Human3.6M S9/S11 motion synthesis | ADE (m): 0.590 -> 0.573 | Improved |
| train DLow on physics-corrected rather than kinematic pose estimates (ablation) | DLow · Human3.6M S9/S11 motion synthesis | FDE (m): 0.698 -> 0.685 | Improved |
| train DLow on physics-corrected rather than kinematic pose estimates (ablation) | DLow · Human3.6M S9/S11 motion synthesis | diversity (APD): 10.53 -> 10.96 | Improved |
| train cVAE on physics-corrected rather than kinematic pose estimates (ablation) | cVAE · Human3.6M S9/S11 motion synthesis | ADE (m): 0.639 -> 0.612 | Improved |
| train cVAE on physics-corrected rather than kinematic pose estimates (ablation) | cVAE · Human3.6M S9/S11 motion synthesis | diversity (APD): 7.419 -> 7.413 | No clear effect |
| add physics loss to kinematic optimization (ablation) | pose optimization · HumanEva walking (15 clips) | body global position error (mm): 190.1 -> 101.1 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- VIBE — VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Trajectory Optimization for Physics-Based Reconstruction of 3D Human Pose From Monocular Video (CVPR 2022 · similarity 0.93)
- Optimal-state Dynamics Estimation for Physics-based Human Motion Capture from Videos (NeurIPS 2024 · similarity 0.91)
- PhyMotion: Structured 3D Motion Reward for Physics-Grounded Human Video Generation (NeurIPS 2026 · similarity 0.89)
- PhysPT: Physics-aware Pretrained Transformer for Estimating Human Dynamics from Monocular Videos (CVPR 2024 · similarity 0.89)
- PhysCtrl: Generative Physics for Controllable and Physics-Grounded Video Generation (NeurIPS 2025 · similarity 0.88)
- HumanVid: Demystifying Training Data for Camera-controllable Human Image Animation (NeurIPS 2024 · 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.
