Research map / 3D vision / Human pose estimation
Leveraging Photometric Consistency Over Time for Sparsely Supervised Hand-Object Reconstruction
Yana Hasson, Bugra Tekin, Federica Bogo, Ivan Laptev, Marc Pollefeys, Cordelia Schmid
Check 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 the proposed direct 6D object-pose regression instead of the baseline approach | hand-object reconstruction · FPHAB action split | object error (mm): 24.9 -> 22.3 | Improved |
| use the proposed hand-pose model instead of the baseline approach | hand-object reconstruction · FPHAB action split | hand error (mm): 15.8 -> 18.0 | Hurt |
| add a skeleton adaptation layer (ablation) | hand reconstruction · FPHAB | hand error (mm): 28.1 -> 27.4 | Improved |
| jointly train hand and object pose estimation rather than train each separately (ablation) | hand-object reconstruction · FPHAB | hand error (mm): 15.7 -> 18.0 | Hurt |
| jointly train hand and object pose estimation rather than train each separately (ablation) | hand-object reconstruction · FPHAB | object error (mm): 21.8 -> 22.3 | Hurt |
| add photometric-consistency self-supervision to sparse-frame training | hand-object reconstruction · HO-3D, below 10% fully annotated frames | pose estimation accuracy | Improved |
| add photometric-consistency self-supervision to training with less than one percent of the data fully supervised | hand-object reconstruction · FPHAB | 2D reprojection error: 7 pixels for objects and 4 pixels for hands; ... 25 and 23% | Improved |
| add photometric-consistency self-supervision with 2.5% of data fully supervised | object pose estimation · FPHAB | 3D object error (mm): 4.7 ... 51% ... 40% | Improved |
| add photometric-consistency self-supervision | object pose estimation · HO-3D | 3D object reconstruction score | No clear effect |
Closest papers by meaning
- HOLD: Category-agnostic 3D Reconstruction of Interacting Hands and Objects from Video (CVPR 2024 · similarity 0.92)
- ComPose: When to Trust Hands for Object Pose Tracking (NeurIPS 2026 · similarity 0.90)
- Hand-held Object Reconstruction from RGB Video with Dynamic Interaction (CVPR 2025 · similarity 0.90)
- BIGS: Bimanual Category-agnostic Interaction Reconstruction from Monocular Videos via 3D Gaussian Splatting (CVPR 2025 · similarity 0.90)
- HandBooster: Boosting 3D Hand-Mesh Reconstruction by Conditional Synthesis and Sampling of Hand-Object Interactions (CVPR 2024 · similarity 0.89)
- EasyHOI: Unleashing the Power of Large Models for Reconstructing Hand-Object Interactions in the Wild (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.
