Research map / Point clouds / Registration and matching
Predator: Registration of 3D Point Clouds With Low Overlap
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 overlap attention to focus interest points on shared regions (ablation) | Predator · 3DLoMatch | Registration Recall (%): 38.9 -> 59.8 | Helped |
| Add overlap scores to the model (ablation) | Predator · 3DLoMatch | Registration Recall (%): 38.9 -> 42.8 | Helped |
| Add cross-overlap scores to overlap scores (ablation) | Predator · 3DLoMatch | Registration Recall (%): 42.8 -> 56.5 | Helped |
| Condition bottleneck features on the other point cloud (ablation) | Predator · 3DLoMatch | Registration Recall (%): 42.8 -> 57.5 | Helped |
| Use probability sampling by overlap × matchability instead of random sampling (ablation) | Predator · 3DLoMatch, 1000 samples | Registration Recall (%): 40.4 -> 62.4 | Helped |
| Use probability sampling by overlap × matchability instead of top-k selection (ablation) | Predator · 3DLoMatch, 1000 samples | Registration Recall (%): 53.1 -> 62.4 | Helped |
| Use overlap × matchability sampling instead of random sampling (ablation) | Predator · ModelLoNet | RRE: 10.985 -> 5.235 | Helped |
| Combine overlap and matchability scores for probability sampling rather than using either score alone (ablation) | Predator · 3DLoMatch | Registration Recall (%): prob. (o): 57.4; prob. (m): 52.2; prob. (om): 59.8 | Helped |
| Add overlap attention to the FCGF backbone (ablation) | FCGF · 3DLoMatch, 250 samples | Registration Recall (%): 26.8 -> 55.9 | Helped |
| Use overlap × matchability probability sampling instead of random sampling (ablation) | Predator · odometryKITTI | RTE [cm]: 8.8 -> 6.8 | Helped |
| Use overlap × matchability probability sampling instead of random sampling (ablation) | Predator · ModelNet | RRE: 2.407 -> 1.739 | Hurt |
Closest papers by meaning
- 3D Focusing-and-Matching Network for Multi-Instance Point Cloud Registration (NeurIPS 2024 · similarity 0.87)
- PointMC: Multi-instance Point Cloud Registration based on Maximal Cliques (ICML 2024 · similarity 0.86)
- Mining and Transferring Feature-Geometry Coherence for Unsupervised Point Cloud Registration (NeurIPS 2024 · similarity 0.86)
- Rectified Point Flow: Generic Point Cloud Pose Estimation (NeurIPS 2025 · similarity 0.86)
- Scalable 3D Registration via Truncated Entry-wise Absolute Residuals (CVPR 2024 · similarity 0.86)
- Generative Point Cloud Registration (ICML 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.