Research map / 3D vision / Human pose estimation
Compositional Human Pose Regression
Xiao Sun, Jiaxiang Shang, Shuang Liang, Yichen Wei
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 2D MPII data to training | Baseline · Human3.6M Protocol 2 (Human3.6M only) | Joint Error: 102.2 -> 64.2 | Improved |
| predict bones with joint-based loss only instead of joints (ablation) | Baseline · Human3.6M Protocol 2 (Human3.6M only) | Joint Error: 102.2 -> 103.3 | Hurt |
| predict bones with joint-based loss only (ablation) | Baseline · Human3.6M Protocol 2 (Human3.6M only) | Bone Error: 65.5 -> 63.5 | Improved |
| predict bones with bone-based loss only (ablation) | Baseline · Human3.6M Protocol 2 (Human3.6M only) | Joint Error: 102.2 -> 104.6 | Hurt |
| predict bones with bone-based loss only (ablation) | Baseline · Human3.6M Protocol 2 (Human3.6M only) | Bone Std: 26.4 -> 21.9 | Improved |
| combine joint and bone loss terms (both) (ablation) | Ours (bone) · Human3.6M Protocol 2 (Human3.6M only) | Joint Error: 104.6 -> 95.2 | Improved |
| add full compositional loss (all) to bone representation (ablation) | Ours (both) · Human3.6M Protocol 2 (Human3.6M only) | Joint Error: 95.2 -> 92.4 | Improved |
| full method (bones + compositional loss) vs baseline | Baseline · Human3.6M Protocol 2 (Human3.6M only) | Joint Error: 102.2 -> 92.4 | Improved |
| full method vs baseline | Baseline · Human3.6M Protocol 2 (Human3.6M only) | Illegal Angle: 3.7% -> 2.5% | Improved |
| full method vs baseline with MPII data | Baseline · Human3.6M Protocol 2 (Human3.6M + MPII) | Joint Error: 64.2 -> 59.1 | Improved |
| add compositional bone representation and loss to two-stage IEF* | IEF* stage 1 · MPII val | PCKH 0.5: 82.9% -> 86.4% | Improved |
| use bone-only loss in IEF* stage 1 (ablation) | IEF* stage 1 · MPII val | PCKH 0.5: 82.9% -> 82.7% | Hurt |
Closest papers by meaning
- Human Pose As Compositional Tokens (CVPR 2023 · similarity 0.90)
- Deeply Learned Compositional Models for Human Pose Estimation (ECCV 2018 · similarity 0.89)
- FlexPose: Pose Distribution Adaptation with Limited Guidance (AAAI 2025 · similarity 0.86)
- Video-Based Human Pose Regression via Decoupled Space-Time Aggregation (CVPR 2024 · similarity 0.86)
- Structure-Aware Correspondence Learning for Relative Pose Estimation (CVPR 2025 · similarity 0.86)
- PoseBH: Prototypical Multi-Dataset Training Beyond Human Pose Estimation (CVPR 2025 · 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.
