Research map / Point clouds / Registration and matching
Register Any Point: Scaling 3D Point Cloud Registration by Flow Matching
Yue Pan, Tao Sun, Liyuan Zhu, Lucas Nunes, Iro Armeni, Jens Behley, Cyrill Stachniss
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 test-time rigidity forcing (ablation) | RAP · Cross-domain multi-view benchmark edge | success rate (%): 83.57 -> 85.92 | Improved |
| add test-time rigidity forcing (ablation) | RAP · Cross-domain multi-view benchmark graph | success rate (%): 72.88 -> 77.31 | Improved |
| add test-time rigidity forcing (ablation) | RAP · 3DLoMatch | success rate (%): 74.63 -> 78.78 | Improved |
| use 10-step Euler integration instead of 1 step (ablation) | RAP · Cross-domain multi-view benchmark graph | success rate (%): 50.32 -> 77.31 | Improved |
| use 10-step Euler integration instead of 1 step (ablation) | RAP · 3DLoMatch | success rate (%): 61.05 -> 78.78 | Improved |
| add local descriptor features as conditions (ablation) | RAP · Cross-domain multi-view benchmark edge | success rate (%): 43.07 -> 85.92 | Improved |
| add local descriptor features as conditions (ablation) | RAP · 3DLoMatch | success rate (%): 50.05 -> 78.78 | Improved |
| increase transformer blocks from 8 to 10 (ablation) | RAP · Cross-domain multi-view benchmark edge | success rate (%): 81.08 -> 85.92 | Improved |
| increase transformer blocks from 6 to 10 (ablation) | RAP · Cross-domain multi-view benchmark edge | success rate (%): 65.64 -> 85.92 | Improved |
| retrain BUFFER-X on RAP's extended training data | BUFFER-X+PGO · Cross-domain multi-view benchmark edge | success rate (%): 50.4 -> 57.7 | Improved |
| replace pairwise+PGO pipeline with single-stage RAP | BUFFER-X*+PGO · Cross-domain multi-view benchmark graph | success rate (%): 36.5 -> 77.3 | Improved |
| replace two-stage pipeline with RAP | BUFFER-X*+PGO · Cross-domain multi-view benchmark | runtime (s): 36.0 -> 8.9 | Improved |
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.
- BUFFER-X — BUFFER-X: Towards Zero-Shot Point Cloud Registration in Diverse Scenes (ICCV 2025) · better on 3 of 3 shared results
- BUFFER — BUFFER: Balancing Accuracy, Efficiency, and Generalizability in Point Cloud Registration (CVPR 2023) · better on 3 of 3 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
- Rectified Point Flow: Generic Point Cloud Pose Estimation (NeurIPS 2025 · similarity 0.92)
- Register Anything Model for Generalizable and Robust Point Cloud Registration (NeurIPS 2026 · similarity 0.92)
- 3D Focusing-and-Matching Network for Multi-Instance Point Cloud Registration (NeurIPS 2024 · similarity 0.90)
- Generative Point Cloud Registration (ICML 2025 · similarity 0.90)
- PointMC: Multi-instance Point Cloud Registration based on Maximal Cliques (ICML 2024 · similarity 0.90)
- Zero-shot RGB-D Point Cloud Registration with Pre-trained Large Vision Model (CVPR 2025 · similarity 0.89)
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.
