Research map / 3D vision / Face and shape
Deep Functional Maps: Structured Prediction for Dense Shape Correspondence
Or Litany, Tal Remez, Emanuele Rodola, Alex Bronstein, Michael Bronstein
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 |
|---|---|---|---|
| train with structured functional-map soft error loss instead of Siamese metric-learning loss (ablation) | FMNet · FAUST templates dense correspondence | geodesic error | Improved |
| use FMNet instead of Chen et al. convex optimization on real scans | FAUST real scans inter-class · FAUST real scans (inter) | average error (cm): 8.30 -> 4.83 | Improved |
| use FMNet instead of Chen et al. on intra-class real scans | FAUST real scans intra-class · FAUST real scans (intra) | average error (cm): 4.86 -> 2.44 | Improved |
| use FMNet instead of Chen et al. on worst error, intra-class | FAUST real scans intra-class · FAUST real scans (intra) | worst error (cm): 26.57 -> 26.16 | No clear effect |
| use the human-specific Stitched Puppet model of Zuffi et al. instead of FMNet | FAUST real scans · FAUST real scans (inter) | average error (cm): 4.83 -> 3.13 | Improved |
| add raw MoNet-style classification loss versus FMNet before refinement | FMNet vs MoNet · FAUST synthetic (before refinement) | accuracy at zero geodesic error | Hurt |
| add correspondence refinement to FMNet (ablation) | FMNet · FAUST synthetic | fraction of matches with zero error | Improved |
| train FMNet on FAUST only and test on SCAPE versus baselines trained on SCAPE | FMNet · SCAPE (transfer from FAUST) | geodesic error | Improved |
Closest papers by meaning
- Deep Geometric Functional Maps: Robust Feature Learning for Shape Correspondence (CVPR 2020 · similarity 0.94)
- Understanding and Improving Features Learned in Deep Functional Maps (CVPR 2023 · similarity 0.92)
- Weakly Supervised Deep Functional Maps for Shape Matching (NeurIPS 2020 · similarity 0.91)
- Weakly-Supervised Learning of Dense Functional Correspondences (ICCV 2025 · similarity 0.91)
- Denoising Functional Maps: Diffusion Models for Shape Correspondence (CVPR 2025 · similarity 0.91)
- Memory-Scalable and Simplified Functional Map Learning (CVPR 2024 · similarity 0.90)
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.
