Research map / Point clouds / Registration and matching
DeepVCP: An End-to-End Deep Neural Network for Point Cloud Registration
Weixin Lu, Guowei Wan, Yao Zhou, Xiangyu Fu, Pengfei Yuan, Shiyu Song
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 the deep feature embedding (DFE) layer before CPG (ablation) | DeepICP · Apollo-SouthBay | mean translation error (m): 0.954 -> 0.018 | Improved |
| add the deep feature embedding (DFE) layer before CPG (ablation) | DeepICP · Apollo-SouthBay | mean angular error (deg): 0.700 -> 0.056 | Improved |
| add FE feature to DFE input (on top of local coordinates and intensity) (ablation) | DeepICP · Apollo-SouthBay | mean translation error (m): 0.024 -> 0.018 | Improved |
| add local coordinates and intensity to DFE input (on top of FE feature) (ablation) | DeepICP · Apollo-SouthBay | mean translation error (m): 0.026 -> 0.018 | Improved |
| add the second network (duplication, 1D z-direction CPG) to the base network (ablation) | DeepICP · Apollo-SouthBay | mean angular error (deg): 0.135 -> 0.056 | Improved |
| add the second network (duplication) on KITTI (ablation) | DeepICP · Apollo-SouthBay | mean angular error (deg): 0.195 -> 0.164 | Improved |
| use the combination of both losses (alpha strictly between 0 and 1) (ablation) | DeepICP · Apollo-SouthBay | mean angular and translation error | Improved |
| vary alpha between 0.1 and 0.9 (ablation) | DeepICP · Apollo-SouthBay | mean angular and translation error | No clear effect |
| replace CPG generated corresponding points with picking the best existing target point (ablation) | DeepICP · Apollo-SouthBay | convergence | Hurt |
Closest papers by meaning
- 3D Point Cloud Registration for Localization Using a Deep Neural Network Auto-Encoder (CVPR 2017 · similarity 0.91)
- DeepPRO: Deep Partial Point Cloud Registration of Objects (ICCV 2021 · similarity 0.91)
- Deep Closest Point: Learning Representations for Point Cloud Registration (ICCV 2019 · similarity 0.91)
- Robust Point Cloud Registration Framework Based on Deep Graph Matching (CVPR 2021 · similarity 0.90)
- 3D Focusing-and-Matching Network for Multi-Instance Point Cloud Registration (NeurIPS 2024 · similarity 0.88)
- PosDiffNet: Positional Neural Diffusion for Point Cloud Registration in a Large Field of View with Perturbations (AAAI 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.
