Research map / 3D vision / Human pose estimation
Global-to-Pixel Regression for Human Mesh Recovery
Yabo Xiao, Mingshu HE, Dongdong Yu
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 Keypoint-Guided Local Encoding Module to global baseline (ablation) | GLNet baseline · 3DPW | PA-MPJPE (mm): 49.6 -> 43.7 | Improved |
| add Dynamic Matching Strategy on top of Local Encoding Module (ablation) | GLNet · 3DPW | MPJPE (mm): 74.3 -> 68.7 | Improved |
| add swing-twist decomposition (ablation) | GLNet · 3DPW | MPJPE (mm): 68.7 -> 66.3 | Improved |
| use sparse 2D keypoint guidance instead of dense segmentation map for local features (ablation) | GLNet · 3DPW | MPJPE (mm): 72.2 -> 66.3 | Improved |
| use sparse 2D keypoint guidance instead of dense IUV map (ablation) | GLNet · 3DPW | MPJPE (mm): 70.9 -> 66.3 | Improved |
| refine only rotation parameters rather than also shape and camera (ablation) | GLNet · 3DPW | MPJPE (mm): 71.2 -> 66.3 | Improved |
| add depth to the matching cost (ablation) | Dynamic Matching Strategy · 3DPW | MPJPE (mm): 66.3 -> 70.4 | Hurt |
| add depth and rotation to the matching cost (ablation) | Dynamic Matching Strategy · 3DPW | MPJPE (mm): 66.3 -> 71.3 | Hurt |
| repeat ground truth T=20 times as positives instead of 1 (ablation) | Dynamic Matching Strategy · 3DPW | MPJPE (mm): 67.4 -> 66.3 | Improved |
| increase positive count T from 20 to 100 (ablation) | Dynamic Matching Strategy · 3DPW | MPJPE (mm): 66.3 -> 68.4 | Hurt |
Closest papers by meaning
- End-to-End Recovery of Human Shape and Pose (CVPR 2018 · similarity 0.92)
- FactorizedHMR: A Hybrid Framework for Video Human Mesh Recovery (NeurIPS 2026 · similarity 0.89)
- ADHMR: Aligning Diffusion-based Human Mesh Recovery via Direct Preference Optimization (ICML 2025 · similarity 0.89)
- MeshPose: Unifying DensePose and 3D Body Mesh Reconstruction (CVPR 2024 · similarity 0.89)
- Incorporating Test-Time Optimization into Training with Dual Networks for Human Mesh Recovery (NeurIPS 2024 · similarity 0.88)
- GenHMR: Generative Human Mesh Recovery (AAAI 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.
