Research map / Label-efficient and robust learning / Recognition and long-tail
Weakly Supervised Deep Functional Maps for Shape Matching
Abhishek Sharma, 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 |
|---|---|---|---|
| remove weak rigid alignment of training data (ablation) | Ours (All losses) · Surreal->Scape | geodesic error x100: 8.3 -> 22 | Hurt |
| remove weak rigid alignment of training data (ablation) | Ours (All losses) · Surreal->Faust | geodesic error x100: 5.0 -> 8.0 | Hurt |
| use only E3 Laplacian commutativity loss instead of all losses (ablation) | Ours · Surreal->Scape | geodesic error x100: 8.3 -> 10.5 | Hurt |
| use only E1 loss instead of all losses (ablation) | Ours · Surreal->Scape | geodesic error x100: 8.3 -> 13 | Hurt |
| use only E2 orthonormality loss instead of all losses (ablation) | Ours · Surreal->Scape | geodesic error x100: 8.3 -> 16 | Hurt |
| use E1+E3 instead of all three losses (ablation) | Ours · Surreal->Faust | geodesic error x100: 5.0 -> 6.3 | Hurt |
| add ZoomOut refinement | Ours · Faust F | geodesic error x100: 3.3 -> 1.9 | Improved |
| add ZoomOut refinement | Ours · Surreal->Scape | geodesic error x100: 8.3 -> 5.5 | Improved |
| weakly supervised method vs fully supervised GeomFmap+zo | Ours + zo · Surreal->Scape | geodesic error x100: 9.2 -> 5.5 | Improved |
| weakly supervised method vs fully supervised GeomFmap+zo | Ours + zo · Surreal->Faust | geodesic error x100: 2.5 -> 2.8 | Hurt |
| train on Scape and test on Faust vs GeomFmap+zo | Ours + zo · S on F | geodesic error x100: 4.3 -> 4.3 | No clear effect |
| use GeomFmap supervised loss with our framework | GeomFmap +Zo · Surreal->Scape | geodesic error x100: 9.2 -> 4.6 | Improved |
Closest papers by meaning
- Understanding and Improving Features Learned in Deep Functional Maps (CVPR 2023 · similarity 0.92)
- Memory-Scalable and Simplified Functional Map Learning (CVPR 2024 · similarity 0.92)
- Denoising Functional Maps: Diffusion Models for Shape Correspondence (CVPR 2025 · similarity 0.89)
- Locality Preserving Refinement for Shape Matching with Functional Maps (AAAI 2024 · similarity 0.88)
- Integrating Efficient Optimal Transport and Functional Maps For Unsupervised Shape Correspondence Learning (CVPR 2024 · similarity 0.87)
- Latent Functional Maps: a spectral framework for representation alignment (NeurIPS 2024 · similarity 0.87)
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.
