Research map / Point clouds / Registration and matching
3D Local Features for Direct Pairwise Registration
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 |
|---|---|---|---|
| Train the features with the full reconstruction, feature-consistency, and pose-prediction objective instead of omitting a loss (ablation) | Ours · Feature matching | fragment matching recall | Helped |
| Add reconstruction loss to feature training (ablation) | Ours · Feature matching | matching performance | Helped |
| Add feature consistency loss to feature training (ablation) | Ours · Feature matching | matching performance | Helped |
| Add pose prediction loss to feature training (ablation) | Ours · Feature matching | matching performance | Helped |
| Use invariant PPF-FoldNet features rather than pose-variant PC-FoldNet features for correspondence matching (ablation) | Ours · 3DMatch feature matching | correspondence inlier ratio | Helped |
| Use direct RelativeNet pose predictions rather than standard RANSAC with the same proposed-feature correspondences | Ours · 3DMatch geometric registration | recall: 0.6901 -> 0.7768 | Helped |
| Keep more mutual nearest-neighbor correspondences by increasing K (ablation) | Ours · 3DMatch geometric registration | recall | Helped |
Closest papers by meaning
- 3D Focusing-and-Matching Network for Multi-Instance Point Cloud Registration (NeurIPS 2024 · similarity 0.89)
- $\pi^2$: A Simple Framework for 2D-to-3D Registration (NeurIPS 2026 · similarity 0.89)
- PointMC: Multi-instance Point Cloud Registration based on Maximal Cliques (ICML 2024 · similarity 0.88)
- Rectified Point Flow: Generic Point Cloud Pose Estimation (NeurIPS 2025 · similarity 0.87)
- Generative Point Cloud Registration (ICML 2025 · similarity 0.87)
- Occlusion-aware Non-Rigid Point Cloud Registration via Unsupervised Neural Deformation Correntropy (ICLR 2025 · 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.