Research map / 3D vision / Face and shape
DiffuMatch: Category-Agnostic Spectral Diffusion Priors for Robust Non-rigid Shape Matching
Emery Pierson, Lei Li, Angela Dai, Maks Ovsjanikov
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 distilled mask, proper loss and SDS on absolute maps instead of vanilla SDS (ablation) | DiffuMatch · SHREC19 (ablation) | geodesic error: 57.3 -> 4.4 | Improved |
| add L_SDS on absolute maps to mask + L_proper (ablation) | DiffuMatch · SHREC19 (ablation) | geodesic error: 6.7 -> 4.4 | Improved |
| add L_proper to mask + L_SDS (ablation) | DiffuMatch · SHREC19 (ablation) | geodesic error: 7.1 -> 4.4 | Improved |
| add distilled mask to L_proper (ablation) | DiffuMatch · SHREC19 (ablation) | geodesic error: 7.7 -> 6.7 | Improved |
| add test-time optimization of feature extractor to mask + Zoomout (ablation) | DiffuMatch · SHREC19 (ablation) | geodesic error: 8.3 -> 4.4 | Improved |
| add axiomatic penalties (orthogonal, bijectivity, Laplacian) (ablation) | DiffuMatch · SHREC19 (ablation) | geodesic error: 4.4 -> 4.3 | No clear effect |
| use distilled diffusion mask instead of Laplacian mask for initialization | Ini + Zoomout · SHREC19 (trained on D-FAUST, zero-shot where applicable) | geodesic error: 13.1 -> 8.3 | Improved |
| use distilled diffusion mask instead of resolvent mask for initialization | Ini + Zoomout · SMAL (trained on D-FAUST, zero-shot where applicable) | geodesic error: 19.1 -> 12.9 | Improved |
| use DiffuMatch instead of SNK | zero-shot matching · SHREC19 (trained on D-FAUST, zero-shot where applicable) | geodesic error: 5.8 -> 3.9 | Improved |
| use DiffuMatch instead of SNK | zero-shot matching · SMAL (trained on D-FAUST, zero-shot where applicable) | geodesic error: 9.1 -> 10.1 | Hurt |
| use DiffuMatch instead of Simplified Fmaps | non-isometric matching · SMAL (trained on D-FAUST, zero-shot where applicable) | geodesic error: 42.1 -> 10.1 | Improved |
| use DiffuMatch instead of Simplified Fmaps | near-isometric matching · SCAPE (trained on D-FAUST, zero-shot where applicable) | geodesic error: 2.3 -> 4.4 | Hurt |
Closest papers by meaning
- Denoising Functional Maps: Diffusion Models for Shape Correspondence (CVPR 2025 · similarity 0.92)
- Memory-Scalable and Simplified Functional Map Learning (CVPR 2024 · similarity 0.91)
- Integrating Efficient Optimal Transport and Functional Maps For Unsupervised Shape Correspondence Learning (CVPR 2024 · similarity 0.89)
- Spectral Meets Spatial: Harmonising 3D Shape Matching and Interpolation (CVPR 2024 · similarity 0.89)
- Hybrid Functional Maps for Crease-Aware Non-Isometric Shape Matching (CVPR 2024 · similarity 0.89)
- Locality Preserving Refinement for Shape Matching with Functional Maps (AAAI 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.
