Research map / Point clouds / Point cloud classification
Learning Compact Geometric Features
Marc Khoury, Qian-Yi Zhou, Vladlen Koltun
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 |
|---|---|---|---|
| replace FPFH with CGF-32 in FGR | FGR · Laser scan test set | pairs correctly aligned (%): 82.96% -> 92.27% | Improved |
| replace FPFH with CGF-32 in FGR | FGR · Laser scan test set | average RMSE (% of diameter): 13.8% -> 9.2% | Improved |
| replace FPFH with CGF-32 in FGR | FGR · SceneNN test set | pairs correctly aligned (%): 88.54% -> 91.19% | Improved |
| replace FPFH with CGF-32 in FGR | FGR · SceneNN test set | average RMSE (cm): 14.86 cm -> 11.83 cm | Improved |
| replace FPFH with CGF-32 in FGR (correctly aligned pairs only) | FGR · SceneNN test set | average RMSE (cm): 4.68 cm -> 4.07 cm | Improved |
| replace FPFH with CGF-32 in FGR | FGR · Redwood benchmark | recall (%): 51.1 -> 60.7 | Improved |
| replace FPFH with CGF-32 in CZK | CZK · Redwood benchmark | recall (%): 59.2 -> 72.0 | Improved |
| replace FPFH with CGF-32 in FGR | FGR · Redwood benchmark | precision (%): 23.2 -> 9.4 | Hurt |
| replace FPFH with CGF-32 in CZK | CZK · Redwood benchmark | precision (%): 19.6 -> 14.6 | Hurt |
| add robust optimization pruning after CZK with CGF-32 | CZK with CGF-32 · Redwood benchmark | precision (%): 14.6 -> 95.1 | Improved |
| add robust optimization pruning after CZK with CGF-32 | CZK with CGF-32 · Redwood benchmark | recall (%): 72.0 -> 71.1 | No clear effect |
| use nonlinear learned embedding instead of PCA into same 32-d space (ablation) | CGF-32 · SceneNN test set | precision within 10 cm (%): 13.4% -> 50.6% | Improved |
Closest papers by meaning
- Fully Convolutional Geometric Features (ICCV 2019 · similarity 0.90)
- Learning general and distinctive 3D local deep descriptors for point cloud registration (TPAMI 2022 · similarity 0.89)
- Mining and Transferring Feature-Geometry Coherence for Unsupervised Point Cloud Registration (NeurIPS 2024 · similarity 0.89)
- Unsupervised Learning of Local Equivariant Descriptors for Point Clouds (TPAMI 2021 · similarity 0.88)
- DH3D: Deep Hierarchical 3D Descriptors for Robust Large-Scale 6DoF Relocalization (ECCV 2020 · similarity 0.88)
- Mining Point Cloud Local Structures by Kernel Correlation and Graph Pooling (CVPR 2018 · 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.
