Research map / 3D vision / Human pose estimation
Generative Modeling of Shape-Dependent Self-Contact Human Poses
Takehiko Ohkawa, Jihyun Lee, Shunsuke Saito, Jason Saragih, Fabian Prada, Yichen Xu, Shoou-I Yu, Ryosuke Furuta, Yoichi Sato, Takaaki Shiratori
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 shape conditioning (ablation) | PAPoseDiff · Goliath-SC generation | FID: 2.18 -> 1.25 | Improved |
| add part-aware self-attention (PASA) (ablation) | PAPoseDiff · Goliath-SC generation | FID: 1.42 -> 1.25 | Improved |
| add shape perturbation (ablation) | PAPoseDiff · Goliath-SC generation | FID: 1.27 -> 1.25 | Improved |
| add shape perturbation (ablation) | PAPoseDiff · Goliath-SC generation | collision ratio: 1.41 -> 1.52 | Hurt |
| add anti-collision guidance (ablation) | PAPoseDiff · Goliath-SC generation | collision ratio: 1.85 -> 1.52 | Improved |
| add anti-collision guidance (ablation) | PAPoseDiff · Goliath-SC generation | FID: 1.28 -> 1.25 | Improved |
| add shape conditioning (ablation) | BUDDI* · Goliath-SC generation | FID: 3.19 -> 2.66 | Improved |
| add shape conditioning (ablation) | VPoser* · Goliath-SC generation | FID: 9.43 -> 9.16 | Improved |
| PAPoseDiff vs BUDDI* (shape-conditional) | generative prior · Goliath-SC generation | FID: 2.66 -> 1.25 | Improved |
| add shape conditioning to refinement prior (ablation) | Hand4Whole + PAPoseDiff · Goliath-SC eval | MPJPE avg (mm): 37.9 -> 35.3 | Improved |
| add shape conditioning to refinement prior (ablation) | HybrIK-X + PAPoseDiff · Goliath-SC eval | MPJPE avg (mm): 45.9 -> 32.4 | Improved |
| add shape conditioning to refinement prior (ablation) | SMPLer-X dagger + PAPoseDiff · Goliath-SC eval | MPJPE avg (mm): 33.7 -> 31.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- HybrIK-X — HybrIK-X: Hybrid Analytical-Neural Inverse Kinematics for Whole-Body Mesh Recovery (TPAMI 2025) · 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
- On Self-Contact and Human Pose (CVPR 2021 · similarity 0.92)
- Learning Complex 3D Human Self-Contact (AAAI 2021 · similarity 0.90)
- imGHUM: Implicit Generative Models of 3D Human Shape and Articulated Pose (ICCV 2021 · similarity 0.87)
- Keep It SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image (ECCV 2016 · similarity 0.87)
- Pose Priors from Language Models (CVPR 2025 · similarity 0.87)
- Estimating Ego-Body Pose from Doubly Sparse Egocentric Video Data (NeurIPS 2024 · similarity 0.86)
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.
