Research map / Point clouds / Registration and matching
ML-SemReg: Boosting Point Cloud Registration with Multi-level Semantic Consistency
Shaocheng Yan, Pengcheng Shi, Jiayuan Li
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 Mask Matching (scene semantic consistency) to the matcher (ablation) | NN matcher + SC2PCR with FPFH · KITTI (NN, SC2PCR, FPFH) | registration recall (%): 54.33 -> 83.44 | Improved |
| add Group Matching on top of Mask Matching (ablation) | NN + SC2PCR with FPFH · KITTI (NN, SC2PCR, FPFH) | registration recall (%): 83.44 -> 85.01 | Improved |
| restrict matching to the same semantic category (MSC) instead of matching groups (ablation) | NN + SC2PCR with FPFH · KITTI (NN, SC2PCR, FPFH) | registration recall (%): 54.33 -> 52.56 | Hurt |
| use Group Matching (LS-Consistency) alone (ablation) | NN + SC2PCR with FPFH · KITTI (NN, SC2PCR, FPFH) | registration recall (%): 54.33 -> 60.90 | Improved |
| add ML-SemReg to Group Matching for GEDI descriptor (ablation) | NN + SC2PCR with GEDI · KITTI (NN, SC2PCR, GEDI) | registration recall (%): 83.77 -> 91.33 | Improved |
| add ML-SemReg to SC2PCR on hard KITTI with FPFH | SC2PCR · KITTI hard (FPFH) | registration recall (%): 8.56 -> 67.82 | Improved |
| set top-K to 1 in Mask Matching (ablation) | ML-SemReg · KITTI (NN, FPFH) | inlier number: 172.01 -> 129.17 | Hurt |
| increase K from 2 to 10 in TL^K (ablation) | ML-SemReg · KITTI (NN, FPFH) | inlier ratio (%): 13.10 -> 2.33 | Hurt |
| add ML-SemReg to RANSAC4M | RANSAC4M · ScanNet 50-60% overlap (FPFH) | registration recall (%): 71.33 -> 68.64 | Hurt |
| add ML-SemReg to SC2PCR on easy KITTI with GEDI | SC2PCR · KITTI easy (GEDI) | translation error (cm): 4.80 -> 5.20 | Hurt |
| swap semantic segmentation model (RangeNet++ to 2DPASS) (ablation) | ML-SemReg · KITTI easy | inlier ratio (%): 29.81 -> 33.03 | No clear effect |
| add ML-SemReg to FPFH matching versus GEDI baseline | NN matcher · KITTI medium | registration recall (%): 89.60 -> 91.01 | Improved |
Closest papers by meaning
- SemReg: Semantics Constrained Point Cloud Registration (ECCV 2024 · similarity 0.93)
- PointMC: Multi-instance Point Cloud Registration based on Maximal Cliques (ICML 2024 · similarity 0.89)
- PSReg: Prior-guided Sparse Mixture of Experts for Point Cloud Registration (AAAI 2025 · similarity 0.89)
- FreeReg: Image-to-Point Cloud Registration Leveraging Pretrained Diffusion Models and Monocular Depth Estimators (ICLR 2024 · similarity 0.89)
- Partial Point Cloud Registration with Multi-view 2D Image Learning (AAAI 2025 · similarity 0.88)
- Mining and Transferring Feature-Geometry Coherence for Unsupervised Point Cloud Registration (NeurIPS 2024 · similarity 0.88)
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.
