Research map / Point clouds / Shape and surface completion
NCP: Neural Correspondence Prior for Effective Unsupervised Shape Matching
Read the paperCheck 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 |
|---|---|---|---|
| train a feature network on the initial noisy maps as supervision | NCP-UN on Smal · Smal test set | mean geodesic error × 100: 7.8 -> 5.8 | Helped |
| use NCP-UN for part label transfer | PartNet · PartNet part label transfer | average IOU (%): 65.8 -> 69.2 | Helped |
| use PointMLP with LIE loss for stage 2 (ablation) | NCP-UN · Chair subset of KeyPointNet | geodesic error (× 100): 9.5 -> 5.1 | Helped |
| use PointMLP with NCE loss for stage 2 (ablation) | NCP-UN · Chair subset of KeyPointNet | geodesic error (× 100): 9.5 -> 5.1 | Helped |
| use PointMLP with FMAP loss for stage 2 (ablation) | NCP-UN · Chair subset of KeyPointNet | geodesic error (× 100): 9.5 -> 5.2 | Helped |
| use DGCNN instead of PointMLP for stage 2 (ablation) | NCP-UN with LIE loss · Chair subset of KeyPointNet | geodesic error (× 100): 9.5 -> 5.0 | Helped |
| use PointNet++ instead of PointMLP for stage 2 (ablation) | NCP-UN with LIE loss · Chair subset of KeyPointNet | geodesic error (× 100): 9.5 -> 5.3 | Helped |
| use ResidualMLP instead of PointMLP for stage 2 (ablation) | NCP-UN with LIE loss · Chair subset of KeyPointNet | geodesic error (× 100): 9.5 -> 8.1 | Helped |
Closest papers by meaning
- NeuroMorph: Unsupervised Shape Interpolation and Correspondence in One Go (CVPR 2021 · similarity 0.91)
- Deep Geometric Functional Maps: Robust Feature Learning for Shape Correspondence (CVPR 2020 · similarity 0.90)
- Deep Shells: Unsupervised Shape Correspondence with Optimal Transport (NeurIPS 2020 · similarity 0.90)
- Unsupervised Learning of Dense Shape Correspondence (CVPR 2019 · similarity 0.89)
- G-MSM: Unsupervised Multi-Shape Matching With Graph-Based Affinity Priors (CVPR 2023 · similarity 0.89)
- Denoising Functional Maps: Diffusion Models for Shape Correspondence (CVPR 2025 · 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.