Research map / 3D vision / Face and shape
Deep Geometric Functional Maps: Robust Feature Learning for Shape Correspondence
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 ZoomOut refinement | Ours · FAUST re-meshed (F) | geodesic error ×100: 3.1 -> 1.9 | Helped |
| Add ZoomOut refinement | Ours · SCAPE re-meshed (S) | geodesic error ×100: 4.4 -> 3.0 | Helped |
| Add ZoomOut refinement | Ours · FAUST re-meshed tested on SCAPE-trained model | geodesic error ×100: 11. -> 9.2 | Helped |
| Add ZoomOut refinement | Ours · SCAPE re-meshed tested on FAUST-trained model | geodesic error ×100: 6.0 -> 4.3 | Helped |
| Use a raw-point-cloud feature extractor instead of FMNet/SHOT descriptors (ablation) | Ours · FAUST re-meshed | geodesic error ×100: 17. -> 3.4; 13. -> 1.9 | Helped |
| Use KPConv instead of PointNet with random sampling (ablation) | Ours · FAUST re-meshed | geodesic error ×100: 18. -> 3.4; 14. -> 1.9 | Helped |
| Add regularization to the functional map layer (ablation) | Ours · FAUST re-meshed | geodesic error ×100: 4.5 -> 3.4 | Helped |
| Add ZoomOut refinement (ablation) | Ours · FAUST re-meshed | geodesic error ×100: 3.4 -> 1.9 | Helped |
| Reduce the SURFMNet functional basis from 120 to 30 eigenvectors (ablation) | SURFMNet · FAUST re-meshed | geodesic error ×100: 0.15 -> 4.5 | Helped |
| Add Laplacian regularization to the functional map layer (ablation) | Ours · FAUST re-meshed | training convergence speed | Helped |
| Add Laplacian regularization to the functional map layer (ablation) | Ours · FAUST re-meshed | map computation robustness | Helped |
| Train with fewer shapes | 3D-CODED · FAUST, SCAPE, and SHREC'19 re-meshed | geodesic error | Hurt |
Closest papers by meaning
- Memory-Scalable and Simplified Functional Map Learning (CVPR 2024 · similarity 0.91)
- Denoising Functional Maps: Diffusion Models for Shape Correspondence (CVPR 2025 · similarity 0.91)
- Spectral Meets Spatial: Harmonising 3D Shape Matching and Interpolation (CVPR 2024 · similarity 0.90)
- Locality Preserving Refinement for Shape Matching with Functional Maps (AAAI 2024 · similarity 0.90)
- Integrating Efficient Optimal Transport and Functional Maps For Unsupervised Shape Correspondence Learning (CVPR 2024 · similarity 0.90)
- Hybrid Functional Maps for Crease-Aware Non-Isometric Shape Matching (CVPR 2024 · similarity 0.89)
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.