Research map / 3D vision / Human pose estimation
SynBody: Synthetic Dataset with Layered Human Models for 3D Human Perception and Modeling
Zhitao Yang, Zhongang Cai, Haiyi Mei, Shuai Liu, Zhaoxi Chen, Weiye Xiao, Yukun Wei, Zhongfei Qing, Chen Wei, Bo Dai, Wayne Wu, Chen Qian and 3 more
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 SynBody to real training data | HMR · 3DPW test | MPJPE: 112.30 -> 95.01 | Improved |
| add SynBody to real training data | SPIN · 3DPW test | MPJPE: 96.90 -> 84.14 | Improved |
| add SynBody to real training data | PARE · 3DPW test | MPJPE: 81.79 -> 78.98 | Improved |
| add SynBody to real training data | VIBE · 3DPW test | MPJPE: 94.88 -> 93.04 | Improved |
| add SynBody to real training data | HMR · AGORA val | MPJPE: 226.71 -> 199.51 | Improved |
| use SynBody-100K instead of AGORA as added synthetic data | HMR · 3DPW test | MPJPE: 101.61 -> 95.28 | Improved |
| use SynBody-100K instead of AGORA as added synthetic data | PARE · 3DPW test | MPJPE: 85.34 -> 79.42 | Improved |
| replace SMPL-XL layered model with SURREAL-style texture-only rendering (ablation) | PARE with SynBody-100K · 3DPW test | MPJPE: 79.4 -> 82.2 | Hurt |
| add SynBody-100K SURREAL-style data to real data (ablation) | PARE · 3DPW test | MPJPE: 81.8 -> 82.2 | No clear effect |
| add SynBody (600K) to real data | OSX · AGORA val SMPL-X | PVE (All): 168.6 -> 155.8 | Improved |
| add SynBody (600K) to real data | OSX · 3DPW test body | MPJPE: 110.2 -> 92.2 | Improved |
| increase amount of SynBody training data (ablation) | HMR trained from scratch · 3DPW test | error | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- SPIN — SPIN: Structure-Preserving Inner Offset Network for Scene Text Recognition (AAAI 2021) · better on 1 of 1 shared result
- VIBE — VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020) · better on 1 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
- BEDLAM: A Synthetic Dataset of Bodies Exhibiting Detailed Lifelike Animated Motion (CVPR 2023 · similarity 0.90)
- SyncHuman: Synchronizing 2D and 3D Generative Models for Single-view Human Reconstruction (NeurIPS 2025 · similarity 0.87)
- GeneMAN: Generalizable Single-Image 3D Human Reconstruction from Multi-Source Human Data (NeurIPS 2025 · similarity 0.87)
- PoseGen: Learning to Generate 3D Human Pose Dataset with NeRF (AAAI 2024 · similarity 0.87)
- MUC: Mixture of Uncalibrated Cameras for Robust 3D Human Body Reconstruction (AAAI 2025 · similarity 0.87)
- SiTH: Single-view Textured Human Reconstruction with Image-Conditioned Diffusion (CVPR 2024 · 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.
