Research map / Point clouds / Shape and surface completion
Voxel-Based Network for Shape Completion by Leveraging Edge Generation
Xiaogang Wang, Marcelo H Ang, Gim Hee Lee
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 multi-scale grid transformation (vs binary voxelization) (ablation) | VE-PCN without EG · Own dataset ablation (2048 points) | CD x1e-4: 5.792 -> 3.600 | Improved |
| use MGT instead of differentiable Gridding (ablation) | VE-PCN without EG · Own dataset ablation (2048 points) | CD x1e-4: 4.768 -> 3.600 | Improved |
| add edge generator (ablation) | VE-PCN · Own dataset ablation (2048 points) | CD x1e-4: 3.600 -> 2.669 | Improved |
| use generated complete edges instead of calculated partial edges (ablation) | VE-PCN · Own dataset ablation (2048 points) | CD x1e-4: 3.586 -> 2.669 | Improved |
| use partial edges instead of no edges (ablation) | VE-PCN without EG · Own dataset ablation (2048 points) | CD x1e-4: 3.600 -> 3.586 | No clear effect |
| add edge generation (ablation) | VE-PCN · KITTI cars | fidelity distance: 0.0279 -> 0.0258 | Improved |
| add Chamfer distance loss (ablation) | VE-PCN · Own dataset loss ablation (2048 points) | CD x1e-4: 3.408 -> 2.669 | Improved |
| add sharper Chamfer distance loss (ablation) | VE-PCN · Own dataset loss ablation (2048 points) | CD x1e-4: 2.886 -> 2.669 | Improved |
| add density loss (ablation) | VE-PCN · Own dataset loss ablation (2048 points) | CD x1e-4: 2.843 -> 2.669 | Improved |
| add BCE loss (ablation) | VE-PCN · Own dataset loss ablation (2048 points) | CD x1e-4: 2.842 -> 2.669 | Improved |
| increase output resolution from 2048 to 16384 points | VE-PCN · Own dataset (seen) | CD x1e-4: 2.669 -> 1.620 | Improved |
| test on unseen categories vs GRNet | VE-PCN · Own dataset unseen | CD x1e-4: 3.892 -> 3.176 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- GRNet — GRNet: Gridding Residual Network for Dense Point Cloud Completion (ECCV 2020) · better on 7 of 7 shared results
- PMP-Net — PMP-Net: Point Cloud Completion by Learning Multi-Step Point Moving Paths (CVPR 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Skeleton-bridged Point Completion: From Global Inference to Local Adjustment (NeurIPS 2020 · similarity 0.90)
- CDPNet: Cross-Modal Dual Phases Network for Point Cloud Completion (AAAI 2024 · similarity 0.90)
- DC-PCN: Point Cloud Completion Network with Dual-Codebook Guided Quantization (AAAI 2025 · similarity 0.89)
- Learning Geometric Transformation for Point Cloud Completion (IJCV 2023 · similarity 0.89)
- SoftPool++: An Encoder–Decoder Network for Point Cloud Completion (IJCV 2022 · similarity 0.89)
- PointAttN: You Only Need Attention for Point Cloud Completion (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.
