Research map / Graph learning / Graph algorithms and clustering
Dense Correspondences between Human Bodies via Learning Transformation Synchronization on Graphs
Xiangru Huang, Haitao Yang, Etienne Vouga, Qixing Huang
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 learned transformation synchronization on top of initial transformations (ablation) | Ours-Desc · SURREAL | average correspondence error (cm): 2.19 -> 1.71 | Improved |
| add learned transformation synchronization on top of initial transformations (ablation) | Ours-Desc · FAUST | average correspondence error (cm): 2.59 -> 1.90 | Improved |
| add learned transformation synchronization on top of initial transformations (ablation) | Ours-Desc · SHREC19 | average correspondence error (cm): 5.84 -> 4.81 | Improved |
| use generic convex optimization instead of learned regularization (ablation) | Ours-Refine · SURREAL | average correspondence error (cm): 1.71 -> 2.13 | Hurt |
| use generic convex optimization instead of learned regularization (ablation) | Ours-Refine · FAUST | average correspondence error (cm): 1.90 -> 2.44 | Hurt |
| use generic convex optimization instead of learned regularization (ablation) | Ours-Refine · SHREC19 | average correspondence error (cm): 4.81 -> 5.77 | Hurt |
| replace learned regularization with ASAP deformation | Ours-Refine · SURREAL | average correspondence error (cm): 1.71 -> 1.93 | Hurt |
| replace learned regularization with ASAP deformation | Ours-Refine · SHREC19 | average correspondence error (cm): 4.81 -> 6.15 | Hurt |
| replace learned regularization with SMPL fitting | Ours-Refine · FAUST | average correspondence error (cm): 1.90 -> 1.98 | Hurt |
| replace Laplacian embedding descriptor with SHOT | Ours-Desc · SURREAL | average correspondence error (cm): 2.19 -> 5.97 | Hurt |
| replace Laplacian embedding descriptor with SHOT | Ours-Desc · SHREC19 | 10cm recall: 0.965 -> 0.840 | Hurt |
| replace Laplacian embedding descriptor with HKS | Ours-Desc · FAUST | average correspondence error (cm): 2.59 -> 9.98 | Hurt |
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.
- ASAP — ASAP: Adaptive Structure Aware Pooling for Learning Hierarchical Graph Representations (AAAI 2020) · better on 9 of 9 shared results
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
- BodyMap: Learning Full-Body Dense Correspondence Map (CVPR 2022 · similarity 0.87)
- DensePose: Dense Human Pose Estimation in the Wild (CVPR 2018 · similarity 0.87)
- DenseReg: Fully Convolutional Dense Shape Regression In-The-Wild (CVPR 2017 · similarity 0.87)
- Dense Human Body Correspondences Using Convolutional Networks (CVPR 2016 · similarity 0.87)
- Learning 3D Human Shape and Pose from Dense Body Parts (TPAMI 2020 · similarity 0.87)
- Denoising Functional Maps: Diffusion Models for Shape Correspondence (CVPR 2025 · similarity 0.85)
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.
