Research map / Point clouds / Registration and matching
RegFormer: An Efficient Projection-Aware Transformer Network for Large-Scale Point Cloud Registration
Jiuming Liu, Guangming Wang, Zhe Liu, Chaokang Jiang, Marc Pollefeys, Hesheng 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 |
|---|---|---|---|
| iteratively refine pose through finer layers instead of outputting from coarsest layer (ablation) | RegFormer · KITTI odometry | Registration recall: 64.2% -> 99.8% | Improved |
| add projection mask to transformer attention (ablation) | RegFormer · KITTI odometry | Registration recall: 98.1% -> 99.8% | Improved |
| use point Swin transformer instead of 2D CNN for feature extraction (ablation) | RegFormer · KITTI odometry | Registration recall: 82.2% -> 99.8% | Improved |
| use point Swin transformer instead of PointNet++ (ablation) | RegFormer · KITTI odometry | Registration recall: 92.4% -> 99.8% | Improved |
| add cross-attention in bijective association transformer (ablation) | RegFormer · KITTI odometry | Registration recall: 98.7% -> 99.8% | Improved |
| add all-to-all point gathering at coarsest layer (ablation) | RegFormer · KITTI odometry | Registration recall: 63.3% -> 99.8% | Improved |
| use BAT instead of cost volume association (ablation) | RegFormer · KITTI odometry | Registration recall: 93.4% -> 99.8% | Improved |
| RegFormer vs RANSAC on KITTI recall and runtime | RegFormer · KITTI odometry | Registration recall: 91.9% -> 99.8% | Improved |
| RegFormer vs Predator on RTE under 10m-apart pairs | RegFormer · KITTI (pairs at least 10m apart) | RTE (cm): 6.8 -> 8.4 | Hurt |
| RegFormer vs HRegNet on NuScenes translation error | RegFormer · NuScenes | RTE (m): RTE 0.18 -> 0.20; RRE 0.45 -> 0.22 | Mixed |
Later papers that compared with this method
1 later accepted paper reports Regformer as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- PointCNN++: Performant Convolution on Native Points (CVPR 2026 · better on 2 of 3 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- HRegNet — HRegNet: A Hierarchical Network for Large-Scale Outdoor LiDAR Point Cloud Registration (ICCV 2021) · better on 3 of 5 shared results
- GeoTransformer — GeoTransformer: Fast and Robust Point Cloud Registration With Geometric Transformer (TPAMI 2023) · better on 1 of 2 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
- A Consistency-Aware Spot-Guided Transformer for Versatile and Hierarchical Point Cloud Registration (NeurIPS 2024 · similarity 0.92)
- Consistency-Aware Spot-Guided Transformer for Accurate and Versatile Point Cloud Registration (TPAMI 2026 · similarity 0.90)
- GeoTransformer: Fast and Robust Point Cloud Registration With Geometric Transformer (TPAMI 2023 · similarity 0.90)
- TULIP: Transformer for Upsampling of LiDAR Point Clouds (CVPR 2024 · similarity 0.88)
- Point Transformer V3: Simpler Faster Stronger (CVPR 2024 · similarity 0.88)
- LiDAR-Net: A Real-scanned 3D Point Cloud Dataset for Indoor Scenes (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.
