Research map / 3D vision / Depth and stereo
UniSH: Unifying Scene and Human Reconstruction in a Feed-Forward Pass
Mengfei Li, Peng Li, Zheng Zhang, Jiahao Lu, Chengfeng Zhao, Wei Xue, Qifeng Liu, Sida Peng, Wenxiao Zhang, Wenhan Luo, Yuan Liu, Yike Guo
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 |
|---|---|---|---|
| Fine-tune π3 on synthetic BEDLAM with original objective (ablation) | π3 · Bonn video depth | Abs Rel: 0.049 -> 0.062 | Hurt |
| Mix BEDLAM with real data in refinement (ablation) | π3 · Bonn video depth | Abs Rel: 0.049 -> 0.051 | Hurt |
| Distill human-centric local loss from MoGe-2 pseudo-depth on unlabeled real data (ablation) | π3 · Bonn video depth | Abs Rel: 0.049 -> 0.035 | Improved |
| Add human-surface refinement (ablation) | π3 · Bonn video depth | delta<1.25: 0.975 -> 0.980 | Improved |
| Supervise scene branch directly for metric scale (no AlignNet scale prediction) (ablation) | UniSH · in-the-wild videos | qualitative | Hurt |
| Add fine-grained alignment on unlabeled real data after coarse synthetic alignment (ablation) | UniSH · in-the-wild SMPL-scene alignment | qualitative | Improved |
| Use UniSH vs GVHMR on global motion | UniSH · EMDB-2 | WA-MPJPE: 111.0 (GVHMR) vs 118.5 (UniSH) | Hurt |
| Use UniSH vs JOSH3R (feed-forward joint) | UniSH · EMDB-2 | WA-MPJPE: 220.0 -> 118.5 | Improved |
| Use UniSH vs GVHMR on RICH | UniSH · RICH | WA-MPJPE: 78.8 vs 118.1 | Hurt |
Later papers that compared with this method
3 later accepted papers report UniSH as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- UniCon3R: Unified Contact-aware 4D Human-Scene Reconstruction from Monocular Video (NeurIPS 2026 · better on 4 of 4 shared results)
- GRAFT: Geometric Refinement and Fitting Transformer for Human Scene Reconstruction (ECCV 2026 · better on 2 of 2 shared results)
- 3R-Adapter: Retrieval, Rewiring, and Refinement for Efficient Adaptation of 3D Reconstruction Model (NeurIPS 2026 · better on 0 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 4 on most shared results.
- TRAM — TRAM: Global Trajectory and Motion of 3D Humans from in-the-wild Videos (ECCV 2024) · better on 1 of 3 shared results
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · better on 2 of 2 shared results
- MonST3R — MonST3R: A Simple Approach for Estimating Geometry in the Presence of Motion (ICLR 2025) · better on 2 of 2 shared results
- DUSt3R — DUSt3R: Geometric 3D Vision Made Easy (CVPR 2024) · better on 2 of 2 shared results
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
- Reconstructing People, Places, and Cameras (CVPR 2025 · similarity 0.92)
- UniCon3R: Unified Contact-aware 4D Human-Scene Reconstruction from Monocular Video (NeurIPS 2026 · similarity 0.91)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.90)
- Synergistic Global-space Camera and Human Reconstruction from Videos (CVPR 2024 · similarity 0.89)
- MultiPly: Reconstruction of Multiple People from Monocular Video in the Wild (CVPR 2024 · similarity 0.88)
- Guess The Unseen: Dynamic 3D Scene Reconstruction from Partial 2D Glimpses (CVPR 2024 · 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.
