Research map / Point clouds / Registration and matching
Learning-Based Point Cloud Registration for 6D Object Pose Estimation in the Real World
Zheng Dang, Lizhou Wang, Yu Guo, Mathieu Salzmann
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 Match Normalization (ablation) | DCP-based registration · TUD-L | ADD (0.1d): 0.23 -> 0.93 | Improved |
| add Match Normalization (ablation) | DCP-based registration · TUD-L | true inlier matches: 10.44 -> 262.96 | Improved |
| add Match Normalization (ablation) | DCP-based registration · TUD-L | predicted matches: 24.84 -> 276.10 | Improved |
| replace SVD loss with NLL loss (ablation) | DCP(v2) · ModelNet40 partial-to-partial (full rotation) | rotation mAP at 5 deg: 0.00 -> 0.35 | Improved |
| replace SVD loss with NLL loss (ablation) | DCP(v2) · ModelNet40 partial-to-partial (45 deg rotation) | rotation mAP at 5 deg: 0.36 -> 0.72 | Improved |
| replace SVD loss with NLL loss (ablation) | DCP(v2) · ModelNet40 partial-to-partial (full rotation) | translation mAP at 0.01: 0.42 -> 0.85 | Improved |
| add Match Normalization, NLL loss and Sinkhorn layer to DCP | DCP(v2) · TUD-L | ADD (0.1d): 0.01 -> 0.85 | Improved |
| add Match Normalization to IDAM | IDAM · TUD-L | ADD (0.1d): 0.05 -> 0.46 | Improved |
| add Match Normalization to IDAM | IDAM · LINEMOD | ADD (0.1d): 0.02 -> 0.11 | Improved |
| add Match Normalization, NLL loss and Sinkhorn layer to DCP | DCP(v2) · LINEMOD | ADD (0.1d): 0.00 -> 0.37 | Improved |
| add Match Normalization, NLL loss and Sinkhorn layer to DCP | DCP(v2) · Occluded-LINEMOD | ADD (0.1d): 0.00 -> 0.28 | Improved |
| add ICP post-processing | Ours-DCP · TUD-L | ADD (0.1d): 0.85 -> 0.93 | Improved |
Closest papers by meaning
- Match Normalization: Learning-Based Point Cloud Registration for 6D Object Pose Estimation in the Real World (TPAMI 2024 · similarity 0.98)
- Learning Symmetry-Aware Geometry Correspondences for 6D Object Pose Estimation (ICCV 2023 · similarity 0.91)
- Center-Based Decoupled Point-cloud Registration for 6D Object Pose Estimation (ICCV 2023 · similarity 0.91)
- Register Any Point: Scaling 3D Point Cloud Registration by Flow Matching (ECCV 2026 · similarity 0.91)
- PointNetLK Revisited (CVPR 2021 · similarity 0.91)
- Zero-shot RGB-D Point Cloud Registration with Pre-trained Large Vision Model (CVPR 2025 · 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.
