Research map / 3D vision / Human pose estimation
Diffusion Models are Efficient Data Generators for Human Mesh Recovery
Yongtao Ge, Wenjia Wang, Yongfan Chen, Fanzhou Wang, Lei Yang, Hao Chen, Chunhua Shen
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 surface normal condition instead of keypoint condition (ablation) | ControlNet · generated humans | mIoU: 41.3 -> 56.8 | Improved |
| use surface normal condition instead of depth condition (ablation) | ControlNet · generated humans | FID: 28.1 -> 24.5 | Improved |
| combine keypoint, depth and normal conditions (ablation) | ControlNet · generated humans | mIoU: 56.8 -> 59.2 | Improved |
| replace ImageNet with COCO keypoint backbone pretraining on BEDLAM+AGORA (ablation) | CLIFF · 3DPW | PA-MPJPE: 51.8 -> 47.4 | Improved |
| replace scratch with COCO pretraining on BEDLAM+AGORA (ablation) | CLIFF · 3DPW | PA-MPJPE: 61.7 -> 47.4 | Improved |
| add HumanWild to BEDLAM+AGORA training (ablation) | CLIFF · 3DPW | PA-MPJPE: 47.4 -> 44.9 | Improved |
| train on HumanWild alone instead of BEDLAM alone | CLIFF · 3DPW | PA-MPJPE: 50.5 -> 51.4 | Hurt |
| add HumanWild to BEDLAM+AGORA+3DPW | CLIFF · 3DPW | PA-MPJPE: 43.0 -> 41.3 | Improved |
| add HumanWild to BEDLAM+AGORA+3DPW | CLIFF · RICH | PVE: 95.6 -> 91.1 | Improved |
| add mixed real-world data R to HumanWild+BEDLAM+AGORA+3DPW (ablation) | CLIFF · 3DPW | PA-MPJPE: 41.3 -> 45.5 | Hurt |
| train on HumanWild instead of PDHuman | Zolly · SPEC-MTP | PA-MPJPE: 102.4 -> 90.0 | Improved |
| add HumanWild to BEDLAM+AGORA+3DPW | CLIFF · MOYO | MPJPE: 143.1 -> 134.6 | Improved |
Closest papers by meaning
- GenHMR: Generative Human Mesh Recovery (AAAI 2025 · similarity 0.90)
- HumanNorm: Learning Normal Diffusion Model for High-quality and Realistic 3D Human Generation (CVPR 2024 · similarity 0.90)
- Human-3Diffusion: Realistic Avatar Creation via Explicit 3D Consistent Diffusion Models (NeurIPS 2024 · similarity 0.89)
- MEAT: Multiview Diffusion Model for Human Generation on Megapixels with Mesh Attention (CVPR 2025 · similarity 0.89)
- PSHuman: Photorealistic Single-image 3D Human Reconstruction using Cross-Scale Multiview Diffusion and Explicit Remeshing (CVPR 2025 · similarity 0.88)
- GeneMAN: Generalizable Single-Image 3D Human Reconstruction from Multi-Source Human Data (NeurIPS 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.
