Research map / Point clouds / Registration and matching
PEAL: Prior-Embedded Explicit Attention Learning for Low-Overlap Point Cloud Registration
Junle Yu, Luwei Ren, Yu Zhang, Wenhui Zhou, Lili Lin, Guojun Dai
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 |
|---|---|---|---|
| Apply PEAL rather than GeoTransformer | Point cloud registration · 3DLoMatch | Registration Recall (%): 75.0 -> 81.7 | Improved |
| Apply PEAL rather than GeoTransformer | Point cloud registration · 3DMatch | Feature Matching Recall (%): 97.9 -> 99.0 | Improved |
| Use non-anchor-to-anchor one-way attention instead of no one-way attention (ablation) | PEAL-3dprior · 3DLoMatch | Registration Recall (%): 74.0 -> 78.8 | Improved |
| Reverse the attention direction to anchor-to-non-anchor (ablation) | PEAL-3dprior · 3DLoMatch | Registration Recall (%): 74.0 -> 16.8 | Hurt |
| Use bidirectional attention (ablation) | PEAL-3dprior · 3DLoMatch | Registration Recall (%): 74.0 -> 71.6 | Hurt |
| Replace explicit attention with prior-embedded implicit attention (ablation) | PEAL-3dprior · 3DLoMatch | Registration Recall (%): 78.8 -> 74.7 | Hurt |
| Select anchor superpoints randomly (ablation) | PEAL-3dprior · 3DLoMatch | Registration Recall (%): 78.8 -> 73.8 | Hurt |
| Restrict anchors to a random 40% of the base region (ablation) | PEAL-3dprior · 3DMatch | Registration Recall (%): 94.1 -> 93.5 | Hurt |
| Expand the anchor region by 16 nearest superpoints (ablation) | PEAL-3dprior · 3DLoMatch | Registration Recall (%): 78.8 -> 77.2 | Hurt |
| Use more image views to obtain 2D overlap prior (ablation) | PEAL-2dprior · 3DLoMatch | Registration Recall (%): 78.8 -> 81.2 | Improved |
| Add iterative transformation updates (ablation) | PEAL-3dprior · 3DLoMatch | Registration Recall (%): 77.8 -> 78.8 | Improved |
| Plug PEAL into Predator | Predator · 3DLoMatch (5000 sampled points) | Registration Recall (%): 59.8 -> 70.2 | Improved |
Later papers that compared with this method
9 later accepted papers report PEAL as a baseline in their results tables; 7 of them report a better number on most of the results they share with it.
- MCI-Net: A Robust Multi-Domain Context Integration Network for Point Cloud Registration (AAAI 2026 · better on 2 of 3 shared results)
- SHReg: Strictly Rotation-Equivariant Point Cloud Registration via Spherical Harmonics (ECCV 2026 · better on 3 of 3 shared results)
- SuP: Sub-cloud Driven Point Cloud Registration (CVPR 2026 · better on 2 of 2 shared results)
- PSReg: Prior-guided Sparse Mixture of Experts for Point Cloud Registration (AAAI 2025 · better on 4 of 4 shared results)
- Partial Point Cloud Registration with Multi-view 2D Image Learning (AAAI 2025 · better on 0 of 4 shared results)
- PARE-Net: Position-Aware Rotation-Equivariant Networks for Robust Point Cloud Registration (ECCV 2024 · better on 4 of 4 shared results)
- ColorPCR: Color Point Cloud Registration with Multi-Stage Geometric-Color Fusion (CVPR 2024 · better on 2 of 2 shared results)
- A Consistency-Aware Spot-Guided Transformer for Versatile and Hierarchical Point Cloud Registration (NeurIPS 2024 · better on 1 of 1 shared result)
1 more not listed.
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- GeoTransformer — GeoTransformer: Fast and Robust Point Cloud Registration With Geometric Transformer (TPAMI 2023) · better on 5 of 5 shared results
- Lepard — Lepard: Learning Partial Point Cloud Matching in Rigid and Deformable Scenes (CVPR 2022) · better on 1 of 1 shared result
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
- PSReg: Prior-guided Sparse Mixture of Experts for Point Cloud Registration (AAAI 2025 · similarity 0.88)
- Dual Focus-Attention Transformer for Robust Point Cloud Registration (CVPR 2025 · similarity 0.87)
- Mining and Transferring Feature-Geometry Coherence for Unsupervised Point Cloud Registration (NeurIPS 2024 · similarity 0.87)
- Implicit Correspondence Learning for Image-to-Point Cloud Registration (CVPR 2025 · similarity 0.87)
- 3D Focusing-and-Matching Network for Multi-Instance Point Cloud Registration (NeurIPS 2024 · similarity 0.86)
- FreeReg: Image-to-Point Cloud Registration Leveraging Pretrained Diffusion Models and Monocular Depth Estimators (ICLR 2024 · 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.
