Research map / Point clouds / Registration and matching
Geometric Transformer for Fast and Robust Point Cloud Registration
Check 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 |
|---|---|---|---|
| Encode superpoint pair distances and triplet angles in geometric self-attention (ablation) | Transformer-based point cloud registration · 3DLoMatch | Registration Recall: 68.4 -> 74.0 | Helped |
| Encode superpoint pair distances and triplet angles in geometric self-attention (ablation) | Transformer-based point cloud registration · 3DMatch | Registration Recall: 89.0 -> 91.5 | Helped |
| Use only pair-wise distances and omit triplet-wise angles (ablation) | Geometric self-attention · 3DLoMatch | Registration Recall: 72.1 -> 74.0 | Hurt |
| Replace geometric structure embedding with absolute coordinate embedding (ablation) | Self-attention for point cloud registration · 3DLoMatch | Registration Recall: 69.3 -> 74.0 | Hurt |
| Replace geometric structure embedding with point pair feature embedding (ablation) | Self-attention for point cloud registration · 3DLoMatch | Registration Recall: 69.5 -> 74.0 | Hurt |
| Replace overlap-aware circle loss with vanilla circle loss (ablation) | GeoTransformer superpoint matching · 3DLoMatch | Registration Recall: 71.5 -> 74.0 | Hurt |
| Replace overlap-aware circle loss with cross-entropy loss (ablation) | GeoTransformer superpoint matching · 3DLoMatch | Registration Recall: 68.4 -> 74.0 | Hurt |
| Replace overlap-aware circle loss with weighted cross-entropy loss (ablation) | GeoTransformer superpoint matching · 3DLoMatch | Registration Recall: 70.7 -> 74.0 | Hurt |
| Apply arbitrary rotations to superpoints (ablation) | Geometric self-attention · 3DLoMatch | Registration Recall: 74.0 -> 73.8 | No clear effect |
| Replace local-to-global registration with weighted SVD (ablation) | GeoTransformer · 3DLoMatch | Registration Recall: 59.9 -> 74.0 | Hurt |
Closest papers by meaning
- Dual Focus-Attention Transformer for Robust Point Cloud Registration (CVPR 2025 · similarity 0.88)
- A Consistency-Aware Spot-Guided Transformer for Versatile and Hierarchical Point Cloud Registration (NeurIPS 2024 · similarity 0.88)
- DV-Matcher: Deformation-based Non-rigid Point Cloud Matching Guided by Pre-trained Visual Features (CVPR 2025 · similarity 0.88)
- PSReg: Prior-guided Sparse Mixture of Experts for Point Cloud Registration (AAAI 2025 · similarity 0.88)
- Dynamic Cues-Assisted Transformer for Robust Point Cloud Registration (CVPR 2024 · similarity 0.88)
- A robust inlier identification algorithm for point cloud registration via $\mathbf{\ell_0}$-minimization (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.