Research map / 3D vision / 3D generation
Reconstructing Hand-Object Interactions in the Wild
Zhe Cao, Ilija Radosavovic, Angjoo Kanazawa, Jitendra Malik
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 interaction loss to individually reconstructed hand and object (ablation) | RHO · In-the-wild images | Hand-object distance: 414.8 -> 71.5 | Improved |
| add interaction loss to individually reconstructed hand and object (ablation) | RHO · In-the-wild images | Collision score: 0 -> 39.8 | Hurt |
| add depth loss after interaction loss (ablation) | RHO · In-the-wild images | Collision score: 39.8 -> 17.6 | Improved |
| add depth loss after interaction loss (ablation) | RHO · In-the-wild images | Hand-object distance: 71.5 -> 75.2 | No clear effect |
| add penetration loss (ablation) | RHO · In-the-wild images | Collision score: 17.6 -> 7.7 | Improved |
| add refinement stage (ablation) | RHO · In-the-wild images | Collision score: 7.7 -> 6.5 | Improved |
| add refinement stage (ablation) | RHO · In-the-wild images | Hand-object distance: 76.4 -> 75.8 | Improved |
| use RHO optimization instead of feed-forward SOTA | RHO · HO3D | Hand MAE (mm): 14.7 -> 9.7 | Improved |
| use RHO optimization instead of feed-forward SOTA | RHO · HO3D | Object Chamfer distance: 26.8 -> 19.9 | Improved |
| use RHO optimization instead of feed-forward SOTA | RHO · FPHA | Hand MAE (mm): 18.0 -> 14.2 | Improved |
| use RHO optimization instead of feed-forward SOTA | RHO · FPHA | Object MAE: 22.3 -> 23.9 | Hurt |
Closest papers by meaning
- EasyHOI: Unleashing the Power of Large Models for Reconstructing Hand-Object Interactions in the Wild (CVPR 2025 · similarity 0.93)
- HOLD: Category-agnostic 3D Reconstruction of Interacting Hands and Objects from Video (CVPR 2024 · similarity 0.93)
- MagicHOI: Leveraging 3D Priors for Accurate Hand-object Reconstruction from Short Monocular Video Clips (ICCV 2025 · similarity 0.92)
- Learning Explicit Contact for Implicit Reconstruction of Hand-Held Objects from Monocular Images (AAAI 2024 · similarity 0.90)
- BIGS: Bimanual Category-agnostic Interaction Reconstruction from Monocular Videos via 3D Gaussian Splatting (CVPR 2025 · similarity 0.90)
- How Do I Do That? Synthesizing 3D Hand Motion and Contacts for Everyday Interactions (CVPR 2025 · similarity 0.89)
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.
