Research map / 3D vision / Human pose estimation
PHD: Personalized 3D Human Body Fitting with Point Diffusion
Hsuan-I Ho, Chen Guo, Po-Chen Wu, Ivan Shugurov, Chengcheng Tang, Abhay Mittal, Sizhe An, Manuel Kaufmann, Linguang Zhang
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 |
|---|---|---|---|
| represent poses as point clouds instead of 6D joint rotations in the prior (ablation) | PointDiT sampling · EMDB1 | PA-MPJPE: 125.2 -> 52.1 | Improved |
| use SHAPify shape instead of per-subject mean shape (ablation) | PHD · EMDB1 | C-MPJPE: 177.3 -> 115.8 | Improved |
| use mean shape instead of zero shape (ablation) | PHD · EMDB1 | C-MPJPE: 170.8 -> 177.3 | Hurt |
| add body measurements to SHAPify (ablation) | SHAPify · EMDB T-pose | Mean joint error: 13.97 -> 11.29 | Improved |
| use ground-truth shape instead of SHAPify | PHD · EMDB1 | Pelvis error: 95.0 -> 84.7 | Improved |
| add point distillation sampling (ablation) | PHD · EMDB1 | MPJPE: 75.8 -> 72.5 | Improved |
| add point distillation sampling (ablation) | PHD · EMDB1 | MVE: 91.7 -> 85.2 | Improved |
| add point distillation sampling (ablation) | PHD · EMDB1 | Pelvis error: 83.3 -> 84.7 | Hurt |
| Point Fitter instead of IK solver (ablation) | PHD · EMDB P1-14 | MVE: 81.9 -> 72.9 | Improved |
| keep all points instead of 25% in Point Fitter (ablation) | PHD · EMDB P1-14 | MVE: 78.1 -> 72.9 | Improved |
| replace ScoreHMR prior with PHD under CameraHMR init (ablation) | CameraHMR · EMDB1 | MPJPE: 74.9 -> 62.5 | Improved |
| replace ScoreHMR prior with PHD under CameraHMR init | CameraHMR · 3DPW | MPJPE: 59.6 -> 59.4 | No clear effect |
Closest papers by meaning
- End-to-End Recovery of Human Shape and Pose (CVPR 2018 · similarity 0.93)
- PC-HMR: Pose Calibration for 3D Human Mesh Recovery from 2D Images/Videos (AAAI 2021 · similarity 0.92)
- TokenHMR: Advancing Human Mesh Recovery with a Tokenized Pose Representation (CVPR 2024 · similarity 0.89)
- AiOS: All-in-One-Stage Expressive Human Pose and Shape Estimation (CVPR 2024 · similarity 0.89)
- Reconstructing Humans with a Biomechanically Accurate Skeleton (CVPR 2025 · similarity 0.89)
- Self-Supervised 3D Human Mesh Recovery from a Single Image with Uncertainty-Aware Learning (AAAI 2024 · similarity 0.89)
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.
