Research map / 3D vision / 3D generation
From 2D Alignment to 3D Plausibility: Unifying Heterogeneous 2D Priors and Penetration-Free Diffusion for Occlusion-Robust Two-Hand Reconstruction
Gaoge Han, Yongkang Cheng, Zhe Chen, Shaoli Huang, Tongliang Liu
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 2D keypoints prior (ablation) | two-hand baseline · InterHand2.6M 5fps test | MPJPE: 7.77 -> 6.48 | Improved |
| add segmentation prior (ablation) | baseline + keypoints · InterHand2.6M 5fps test | MPJPE: 6.48 -> 6.19 | Improved |
| add depth prior (ablation) | baseline + keypoints + segmentation · InterHand2.6M 5fps test | MRRPE: 24.52 -> 22.38 | Improved |
| add depth prior (ablation) | baseline + keypoints + segmentation · InterHand2.6M 5fps test | MPJPE-Z: 4.40 -> 3.37 | Improved |
| add penetration-free diffusion (ablation) | FAE model · InterHand2.6M 5fps test | MPJPE: 5.74 -> 5.36 | Improved |
| add penetration-free diffusion (ablation) | FAE model · InterHand2.6M 5fps test | MRRPE: 22.38 -> 21.60 | Improved |
| add all priors and diffusion (ablation) | two-hand baseline · InterHand2.6M 5fps test | MRRPE: 25.30 -> 21.60 | Improved |
| use fusion alignment encoder instead of foundation encoder at inference (ablation) | two-hand model · RTX 3090 | FPS: 3 -> 56 | Improved |
| use fusion alignment encoder instead of foundation encoder at inference (ablation) | two-hand model · InterHand2.6M 5fps test | MPJPE: 5.54 -> 5.74 | Hurt |
| use proposed penetration-free diffusion instead of InterHandGen | two-hand reconstruction · InterHand2.6M | PenVol: 0.76 -> 0.11 | Improved |
| use Ours instead of 4DHands | two-hand reconstruction · InterHand2.6M 5fps test | MPJPE: 7.49 -> 5.36 | Improved |
| use Ours instead of 4DHands on in-the-wild HIC | two-hand reconstruction · HIC | MPJPE: 9.32 -> 6.67 | Improved |
Closest papers by meaning
- InterHandGen: Two-Hand Interaction Generation via Cascaded Reverse Diffusion (CVPR 2024 · similarity 0.90)
- HandBooster: Boosting 3D Hand-Mesh Reconstruction by Conditional Synthesis and Sampling of Hand-Object Interactions (CVPR 2024 · similarity 0.88)
- Learning Explicit Contact for Implicit Reconstruction of Hand-Held Objects from Monocular Images (AAAI 2024 · similarity 0.87)
- BIGS: Bimanual Category-agnostic Interaction Reconstruction from Monocular Videos via 3D Gaussian Splatting (CVPR 2025 · similarity 0.87)
- Single-view Image to Novel-view Generation for Hand-Object Interactions (AAAI 2025 · similarity 0.87)
- Hand-held Object Reconstruction from RGB Video with Dynamic Interaction (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.
