Research map / Point clouds / Registration and matching
SIRA-PCR: Sim-to-Real Adaptation for 3D Point Cloud Registration
Suyi Chen, Hao Xu, Ru Li, Guanghui Liu, Chi-Wing Fu, Shuaicheng Liu
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 |
|---|---|---|---|
| insert ShapeNet objects into 3D-FRONT scenes (ablation) | FlyingShapes · 3DLoMatch | Registration Recall: 33.5 -> 55.1 | Improved |
| delete flat planes with 50% probability (ablation) | FlyingShapes · 3DLoMatch | Registration Recall: 55.1 -> 57.4 | Improved |
| add Structured3D to FlyingShapes (ablation) | FlyingShapes · 3DLoMatch | Registration Recall: 57.4 -> 63.8 | Improved |
| add uniform or Gaussian noise to synthetic data (ablation) | GeoTrans pretrained on FlyingShapes · 3DLoMatch | Registration Recall: 63.8 -> 62.7 (uniform), 63.8 (Gaussian) | No clear effect |
| apply TSDF fusion fine-tuning (ablation) | GeoTrans pretrained on FlyingShapes · 3DLoMatch | Registration Recall: 63.8 -> 63.9 | No clear effect |
| add SIRA sim-to-real adaptation (ablation) | GeoTrans pretrained on FlyingShapes · 3DLoMatch | Registration Recall: 63.8 -> 67.4 | Improved |
| add adaptive re-sample module (ARM) (ablation) | SIRA · 3DLoMatch | Registration Recall: 65.3 -> 67.4 | Improved |
| add multi-scale discriminator (ablation) | SIRA · 3DLoMatch | Registration Recall: 66.6 -> 67.4 | Improved |
| fine-tune synthetic-trained model on 3DMatch | SIRA-PCR · ETH (RANSAC, 250 samples) | Registration Recall: 95.4 -> 82.2 | Hurt |
| train only on synthetic data (SIRA-PCR) versus GeoTrans trained on real data | GeoTrans · 3DLoMatch | Registration Recall: 74.0 -> 67.4 | Hurt |
| fine-tune SIRA-PCR on 3DMatch, using RANSAC | SIRA-PCR · 3DLoMatch RR (5000 samples) | Registration Recall: GeoTrans 75.0 -> SIRA-PCR 73.5 | Mixed |
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.
- RegTR — REGTR: End-to-End Point Cloud Correspondences With Transformers (CVPR 2022) · better on 1 of 1 shared result
- CoFiNet — CoFiNet: Reliable Coarse-to-fine Correspondences for Robust PointCloud Registration (NeurIPS 2021) · 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
- Point-TTA: Test-Time Adaptation for Point Cloud Registration Using Multitask Meta-Auxiliary Learning (ICCV 2023 · similarity 0.90)
- Deep Learning-Based Point Cloud Registration: A Comprehensive Survey and Taxonomy (IJCV 2026 · similarity 0.89)
- Cross-PCR: A Robust Cross-Source Point Cloud Registration Framework (AAAI 2025 · similarity 0.86)
- SMART-PC: Skeletal Model Adaptation for Robust Test-Time Training in Point Clouds (ICML 2025 · similarity 0.86)
- Generative Point Cloud Registration (ICML 2025 · similarity 0.86)
- MICAS: Multi-grained In-Context Adaptive Sampling for 3D Point Cloud Processing (CVPR 2025 · similarity 0.85)
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.
