Research map / 3D vision / 3D generation
Recovering Physically Plausible Human-Object Interactions from Monocular Videos
Dingbang Huang, Etienne Vouga, Qixing Huang, Georgios Pavlakos
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 adaptive sampling to naive training (ablation) | physics-based HOI tracker · BEHAVE | SR-B: 11.4 -> 14.3 | Improved |
| add adaptive sampling to naive training (ablation) | physics-based HOI tracker · BEHAVE | SR-F: 24.5 -> 40.4 | Improved |
| add kinematic updates as tracking targets (ablation) | adaptive sampling tracker · BEHAVE | SR-B: 14.3 -> 17.1 | Improved |
| add kinematic updates as tracking targets (ablation) | adaptive sampling tracker · BEHAVE | SR-F: 40.4 -> 43.5 | Improved |
| add dual propagation (forward and backward policies) (ablation) | adaptive sampling + kinematic updates · BEHAVE | SR-B: 17.1 -> 51.4 | Improved |
| add dual propagation (forward and backward policies) (ablation) | adaptive sampling + kinematic updates · BEHAVE | SR-F: 43.5 -> 60.0 | Improved |
| use kinematic updates as tracking target, not only for initialization (ablation) | full method · BEHAVE | SR-B: 40.0 -> 51.4 | Improved |
| use kinematic updates as tracking target, not only for initialization (ablation) | full method · BEHAVE | SR-F: 59.4 -> 60.0 | Improved |
| use two policies instead of one in dual propagation (ablation) | full method · BEHAVE | SR-B: 48.5 -> 51.4 | Improved |
| use two policies instead of one in dual propagation (ablation) | full method · BEHAVE | SR-F: 59.7 -> 60.0 | No clear effect |
| physics-based refinement over kinematic VisTracker | VisTracker · BEHAVE | ContRate-h: 0.52 -> 0.89 | Improved |
| physics-based refinement over kinematic VisTracker | VisTracker · BEHAVE | CD-H: 5.39 -> 6.82 | Hurt |
Closest papers by meaning
- Real2Sim in HOI: Toward Physically Plausible HOI Reconstruction from Monocular Videos (NeurIPS 2026 · similarity 0.92)
- InterMimic: Towards Universal Whole-Body Control for Physics-Based Human-Object Interactions (CVPR 2025 · similarity 0.89)
- Closely Interactive Human Reconstruction with Proxemics and Physics-Guided Adaption (CVPR 2024 · similarity 0.88)
- CG-HOI: Contact-Guided 3D Human-Object Interaction Generation (CVPR 2024 · similarity 0.88)
- UniCon3R: Unified Contact-aware 4D Human-Scene Reconstruction from Monocular Video (NeurIPS 2026 · similarity 0.88)
- Reconstructing Close Human Interaction with Appearance and Proxemics Reasoning (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.
