Research map / Point clouds / Point cloud classification
Equi-GSPR: Equivariant SE(3) Graph Network Model for Sparse Point Cloud Registration
Xueyang Kang, Zhaoliang Luan, Kourosh Khoshelham, Bing WANG
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 |
|---|---|---|---|
| add learned descriptor layers (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 61.39 -> 94.60 | Improved |
| add equivariant graph layers (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 62.45 -> 94.60 | Improved |
| use equivariant layers instead of normal GCNN (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 68.52 -> 94.60 | Improved |
| add Low-Rank Feature Transformation (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 83.09 -> 94.60 | Improved |
| add rank regularizer (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 76.45 -> 94.60 | Improved |
| add sub-matrix rank verification (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 87.76 -> 94.60 | Improved |
| use ball query instead of kNN graph construction (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 92.37 -> 94.60 | Improved |
| use learned PointNet++ descriptor instead of pre-trained FCGF (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 93.87 -> 94.60 | Improved |
| use learned descriptor instead of hand-crafted FPHF (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 83.62 -> 94.60 | Improved |
| replace descriptor layers with SpinNet equivariant descriptor (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 94.60 -> 82.16 | Hurt |
| increase sampled points from 1024 to 2048 or 4096 (ablation) | Equi-GSPR · 3DMatch | Registration Recall (%): 94.6 -> 94.8 (2048), 95.3 (4096) | No clear effect |
| add equivariant layers (latency cost) (ablation) | Equi-GSPR · 3DMatch | Time (s): 0.06 -> 0.12 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SpinNet — SpinNet: Learning a General Surface Descriptor for 3D Point Cloud Registration (CVPR 2021) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Equivariant Point Network for 3D Point Cloud Analysis (CVPR 2021 · similarity 0.92)
- E2PN: Efficient SE(3)-Equivariant Point Network (CVPR 2023 · similarity 0.90)
- Correspondence-Free SE(3) Point Cloud Registration in RKHS via Unsupervised Equivariant Learning (ECCV 2024 · similarity 0.89)
- Approximately Piecewise E(3) Equivariant Point Networks (ICLR 2024 · similarity 0.87)
- SE(3)-bi-equivariant Transformers for Point Cloud Assembly (NeurIPS 2024 · similarity 0.87)
- CARL: A Framework for Equivariant Image Registration (CVPR 2025 · similarity 0.86)
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.
