Research map / 3D vision / Human pose estimation
Shape-Aware Human Pose and Shape Reconstruction Using Multi-View Images
Junbang Liang, Ming C. Lin
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 |
|---|---|---|---|
| use multi-view input instead of single view (ablation) | Ours · Human3.6M (Procrustes aligned, with synthetic training) | MPJPE: 58.55 -> 45.13 | Improved |
| use multi-view input instead of single view (ablation) | Ours · MPI_INF_3DHP validation (with synthetic training) | MPJPE: 84 -> 62 | Improved |
| use multi-view input instead of single view (ablation) | Ours · Synthetic dataset (with synthetic training) | Hausdorff distance: 65 -> 53 | Improved |
| add synthetic shape-aware dataset to training (ablation) | HMR · Synthetic dataset | Hausdorff distance: 208 -> 83 | Improved |
| add synthetic shape-aware dataset to training (ablation) | Ours (single) · Synthetic dataset | Hausdorff distance: 283 -> 65 | Improved |
| add synthetic dataset to training (ablation) | Ours (single) · Human3.6M (Procrustes aligned) | MPJPE: 59.09 -> 58.55 | No clear effect |
| add synthetic dataset to training (ablation) | Ours (multi) · MPI_INF_3DHP validation | MPJPE: 59 -> 62 | No clear effect |
| use multi-view model instead of HMR for shape from tape measurements | Ours (multi) vs HMR · Tape-measured body measurements (sitting) | average relative error (%): 7.29% -> 5.26% | Improved |
| use single-view model instead of HMR for tape-measured sitting shape | Ours (single) vs HMR · Tape-measured body measurements (sitting) | average relative error (%): 7.29% -> 10.18% | Hurt |
| use Procrustes alignment on multi-view prediction | Ours · Human3.6M multi-view comparison | MPJPE: 79.85 -> 45.13 | Improved |
| use refined SMPL shape model instead of Alldieck et al. | Ours vs Alldieck et al. · 3D People in the Wild | mean vertex error (GT pose): 47.07 -> 20.96 | Improved |
Closest papers by meaning
- 3D Multi-bodies: Fitting Sets of Plausible 3D Human Models to Ambiguous Image Data (NeurIPS 2020 · similarity 0.90)
- Skeleton-Aware 3D Human Shape Reconstruction From Point Clouds (ICCV 2019 · similarity 0.89)
- Learning To Regress Bodies From Images Using Differentiable Semantic Rendering (ICCV 2021 · similarity 0.89)
- MultiGO: Towards Multi-level Geometry Learning for Monocular 3D Textured Human Reconstruction (CVPR 2025 · similarity 0.89)
- MUC: Mixture of Uncalibrated Cameras for Robust 3D Human Body Reconstruction (AAAI 2025 · similarity 0.88)
- SCULPT: Shape-Conditioned Unpaired Learning of Pose-dependent Clothed and Textured Human Meshes (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.
