Research map / Point clouds / Registration and matching
Deep Graph-Based Spatial Consistency for Robust Non-Rigid Point Cloud Registration
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 |
|---|---|---|---|
| apply GraphSCNet outlier pruning to Lepard correspondences | Lepard · 4DMatch | AccS: 54.2 -> 70.1 | Helped |
| apply GraphSCNet outlier pruning to Lepard correspondences | Lepard · 4DLoMatch | AccS: 17.4 -> 40.0 | Helped |
| apply GraphSCNet outlier pruning to GeoTransformer correspondences | GeoTransformer · 4DMatch | AccS: 65.5 -> 72.3 | Helped |
| apply GraphSCNet outlier pruning to GeoTransformer correspondences | GeoTransformer · 4DLoMatch | AccS: 31.4 -> 41.0 | Helped |
| use node coverage σn = 0.04 instead of 0.08 (ablation) | GraphSCNet · 4DMatch | AccS: 72.3 -> 70.6 | Hurt |
| increase node coverage to σn = 0.32 (ablation) | GraphSCNet · 4DMatch | AccS: 72.3 -> 69.5 | Hurt |
| replace furthest-point node sampling with random sampling (ablation) | GraphSCNet · 4DMatch | AccS: 72.3 -> 72.5 | No clear effect |
| use correspondence-centered kNN graph construction (ablation) | GraphSCNet · 4DMatch | AccS: 72.3 -> 71.9 | Hurt |
| assign correspondences to only one neighboring node (ablation) | GraphSCNet · 4DMatch | AccS: 72.3 -> 59.7 | Hurt |
| assign correspondences to nine neighboring nodes (ablation) | GraphSCNet · 4DMatch | AccS: 72.3 -> 69.8 | Hurt |
| replace Euclidean graph distance with geodesic distance (ablation) | GraphSCNet · 4DMatch | AccS: 72.3 -> 71.1 | Hurt |
| remove Fourier positional encoding (ablation) | GraphSCNet · 4DMatch | AccS: 72.3 -> 72.3 | No clear effect |
Closest papers by meaning
- Multi-scale Consistency for Robust 3D Registration via Hierarchical Sinkhorn Tree (NeurIPS 2024 · similarity 0.90)
- Occlusion-aware Non-Rigid Point Cloud Registration via Unsupervised Neural Deformation Correntropy (ICLR 2025 · similarity 0.89)
- PointMC: Multi-instance Point Cloud Registration based on Maximal Cliques (ICML 2024 · similarity 0.89)
- A robust inlier identification algorithm for point cloud registration via $\mathbf{\ell_0}$-minimization (NeurIPS 2024 · similarity 0.88)
- Scalable 3D Registration via Truncated Entry-wise Absolute Residuals (CVPR 2024 · similarity 0.87)
- Inlier Confidence Calibration for Point Cloud Registration (CVPR 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.