Research map / 3D vision / Face and shape
3D Reconstruction of Objects in Hands without Real World 3D Supervision
Aditya Prakash, Matthew Chang, Matthew Jin, Ruisen Tu, Saurabh Gupta
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 multiview 2D mask supervision and shape priors instead of 3D lab supervision | AC-SDF / AC-OCC · MOW object generalization | F-score@5mm: 0.108 -> 0.121 | Improved |
| add HO3D and HOI4D lab 3D data to AC-SDF training (ablation) | AC-SDF · MOW object generalization | F-score@5mm: 0.108 -> 0.095 | Hurt |
| add multiview occupancy loss L_occ (ablation) | HORSE · MOW object generalization | Chamfer distance (mm): 8.69 -> 7.26 | Improved |
| add consistency loss L_consis alone (ablation) | HORSE · MOW object generalization | Chamfer distance (mm): 8.69 -> 12.75 | Hurt |
| add L_consis on top of L_occ (ablation) | HORSE · MOW object generalization | Chamfer distance (mm): 7.26 -> 6.93 | Improved |
| add shape prior loss L_shape (ablation) | HORSE · MOW object generalization | Chamfer distance (mm): 6.93 -> 6.76 | Improved |
| add automated filtering of frames with inaccurate hand poses (ablation) | HORSE · HO3D view generalization | F-score@5mm: 0.213 -> 0.234 | Improved |
| use ground truth hand pose instead of predicted (ablation) | HORSE · HO3D view generalization | F-score@5mm: 0.234 -> 0.243 | No clear effect |
| use mask-guided rejection sampling instead of uniform sampling (ablation) | HORSE · MOW object generalization | Chamfer distance (mm): 10.29 -> 6.93 | Improved |
| add negative points from hand masks (ablation) | HORSE mask-guided sampling · MOW object generalization | Chamfer distance (mm): 7.69 -> 6.93 | Improved |
| use 2D slice discriminator instead of 3D volume discriminator (ablation) | HORSE · MOW object generalization | Chamfer distance (mm): 7.83 -> 6.76 | Improved |
| add 2D-slice discriminator (ablation) | HORSE · MOW object generalization | F-score@5mm: 0.117 -> 0.121 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- gSDF — GSDF: 3DGS Meets SDF for Improved Neural Rendering and Reconstruction (NeurIPS 2024) · better on 1 of 1 shared result
- DDFHO — DDF-HO: Hand-Held Object Reconstruction via Conditional Directed Distance Field (NeurIPS 2023) · 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
- In-Hand 3D Object Reconstruction from a Monocular RGB Video (AAAI 2024 · similarity 0.91)
- HOLD: Category-agnostic 3D Reconstruction of Interacting Hands and Objects from Video (CVPR 2024 · similarity 0.91)
- Hand-held Object Reconstruction from RGB Video with Dynamic Interaction (CVPR 2025 · similarity 0.89)
- Learning Explicit Contact for Implicit Reconstruction of Hand-Held Objects from Monocular Images (AAAI 2024 · similarity 0.89)
- HandBooster: Boosting 3D Hand-Mesh Reconstruction by Conditional Synthesis and Sampling of Hand-Object Interactions (CVPR 2024 · similarity 0.88)
- EasyHOI: Unleashing the Power of Large Models for Reconstructing Hand-Object Interactions in the Wild (CVPR 2025 · similarity 0.88)
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.
