Research map / 3D vision / Human pose estimation
Learning Visibility for Robust Dense Human Body Estimation
Chun-Han Yao, Jimei Yang, Duygu Ceylan, Yi Zhou, Yang Zhou, Ming-Hsuan Yang
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 truncation modeling L_vis^{x,y} (ablation) | VisDB mesh · 3DPW | MPVE: 91.1 -> 85.5 | Improved |
| add occlusion modeling L_vis^z (ablation) | VisDB mesh · 3DPW | MPVE: 88.0 -> 85.5 | Improved |
| add depth ordering loss L_depth (ablation) | VisDB mesh · 3DPW | MPVE: 88.9 -> 85.5 | Improved |
| add UV correspondence loss L_uv (ablation) | VisDB mesh · 3DPW | MPVE: 87.1 -> 85.5 | Improved |
| add all four visibility/depth/UV components (ablation) | I2L-MeshNet baseline · 3DPW | MPJPE: 84.5 -> 73.5 | Improved |
| use dense visibility weighting in SMPL regression (ablation) | VisDB SMPL · 3DPW | MPVE: drops by 8.7mm | Improved |
| add test-time optimization with visibility weighting (ablation) | VisDB SMPL regression · 3DPW | MPVE: further reduce the error by 3.8mm | Improved |
| add optimization without visibility to regression without visibility (ablation) | VisDB SMPL · 3DPW | MPVE: 96.2 -> 93.9 | Improved |
| visibility-aware VisDB over retrained I2L-MeshNet baseline | I2L-MeshNet · 3DPW-OCC | MPVE: 120.2 -> 114.0 | Improved |
| visibility-aware VisDB over PARE | PARE · 3DPW | MPVE: 94.2 -> 83.5 | Improved |
| SMPL parameter output instead of mesh output | VisDB · 3DPW | MPJPE: 73.5 -> 72.1 | Improved |
| VisDB (mesh) over Mesh Graphormer | Mesh Graphormer · Human3.6M | PA-MPJPE: 34.5 -> 34.5 | No clear effect |
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.
- I2L-MeshNet — I2L-MeshNet: Image-to-Lixel Prediction Network for Accurate 3D Human Pose and Mesh Estimation from a Single RGB Image (ECCV 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
- DenseRaC: Joint 3D Pose and Shape Estimation by Dense Render-and-Compare (ICCV 2019 · similarity 0.89)
- DensePose: Dense Human Pose Estimation in the Wild (CVPR 2018 · similarity 0.89)
- BodyMap: Learning Full-Body Dense Correspondence Map (CVPR 2022 · similarity 0.89)
- SimPose: Effectively Learning DensePose and Surface Normals of People from Simulated Data (ECCV 2020 · similarity 0.89)
- Divide and Fuse: Body Part Mesh Recovery from Partially Visible Human Images (ECCV 2024 · similarity 0.89)
- Learning To Regress Bodies From Images Using Differentiable Semantic Rendering (ICCV 2021 · 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.
