Research map / 3D vision / Human pose estimation
Spatiotemporal Bundle Adjustment for Dynamic 3D Human Reconstruction in the Wild
Minh Vo, Yaser Ajmal Sheikh, Srinivasa G. Narasimhan
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 motion prior instead of geometric triangulation | spatiotemporal BA · CMU Mocap synthetic, dynamic points | reprojection error: 17.8 -> 0.85 | Improved |
| use motion prior instead of generic cubic B-spline trajectory basis | spatiotemporal BA · CMU Mocap synthetic, dynamic points | reprojection error: 1.68 -> 0.85 | Improved |
| add DCT resampling to motion prior (incremental) (ablation) | spatiotemporal BA · CMU Mocap synthetic, dynamic points | reprojection error: 0.85 -> 0.74 | Improved |
| use divide and conquer instead of incremental reconstruction | spatiotemporal BA · CMU Mocap synthetic | reprojection error and speed: 0.85 -> 0.89 | No clear effect |
| add DCT resampling instead of direct motion prior (ablation) | spatiotemporal BA · CMU Mocap synthetic | mean 3D error (cm): 6.6 -> 6.5 | No clear effect |
| use least action prior instead of least kinetic energy prior | motion prior · CMU Mocap time offset estimation | offset bias | Hurt |
| use least force prior instead of least kinetic energy prior at low cumulative frame rate | motion prior · CMU Mocap time offset estimation | sequencing expressiveness | Improved |
| use least force prior instead of least kinetic at high cumulative frame rate (>=120fps) | motion prior · CMU Mocap time offset estimation | alignment result | No clear effect |
| model rolling shutter readout speed gamma (ablation) | spatiotemporal BA · Checkerboard dynamic points | reprojection error: 1.52 -> 1.21 | Improved |
| model rolling shutter readout speed gamma (ablation) | spatiotemporal BA · Dance dynamic points | reprojection error: 2.38 -> 2.13 | Improved |
| use motion prior instead of geometric triangulation | spatiotemporal BA · Dance real video dynamic points | RMSE (pixels): 5.23 -> 2.12 | Improved |
| add resampling to motion prior (ablation) | spatiotemporal BA · real videos dynamic points | RMSE (pixels): 1.21 -> 1.15 | Improved |
Closest papers by meaning
- HumanMM: Global Human Motion Recovery from Multi-shot Videos (CVPR 2025 · similarity 0.88)
- Link to the Past: Temporal Propagation for Fast 3D Human Reconstruction from Monocular Video (CVPR 2025 · similarity 0.88)
- 4D3R: Motion-Aware Neural Reconstruction and Rendering of Dynamic Scenes from Monocular Videos (NeurIPS 2025 · similarity 0.88)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.87)
- Enhancing 3D Reconstruction for Dynamic Scenes (NeurIPS 2025 · similarity 0.87)
- Reconstructing People, Places, and Cameras (CVPR 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.
