Research map / Point clouds / Shape and surface completion
SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution With Skip-Transformer
Peng Xiang, Xin Wen, Yu-Shen Liu, Yan-Pei Cao, Pengfei Wan, Wen Zheng, Zhizhong Han
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 skip-transformer attention instead of self-attention (ablation) | SnowflakeNet · Completion3D val | average L2 Chamfer distance (lower better): 8.89 -> 8.48 | Improved |
| add transformer attention in the skip connection instead of plain feature addition (ablation) | SnowflakeNet · Completion3D val | average L2 Chamfer distance (lower better): 9.30 -> 8.48 | Improved |
| add skip-transformer between SPD layers (ablation) | SnowflakeNet · Completion3D val | average L2 Chamfer distance (lower better): 9.39 -> 8.48 | Improved |
| use self-attention instead of no attention in skip connection (ablation) | SnowflakeNet · Completion3D val | average L2 Chamfer distance (lower better): 9.30 -> 8.89 | Improved |
| use point-wise splitting instead of folding-based feature expansion (ablation) | SnowflakeNet · Completion3D val | average L2 Chamfer distance (lower better): 8.80 -> 8.48 | Improved |
| add partial matching loss (ablation) | SnowflakeNet · Completion3D val | average L2 Chamfer distance (lower better): 8.50 -> 8.48 | Improved |
| replace PCN decoder with SPD plus skip-transformer decoder (ablation) | PCN encoder · Completion3D val | average L2 Chamfer distance (lower better): 13.30 -> 8.93 | Improved |
| use SnowflakeNet instead of NSFA | PCN dataset · PCN dataset | average L1 Chamfer distance x1e3: 8.06 -> 7.21 | Improved |
| use SnowflakeNet instead of PMP-Net | Completion3D test · Completion3D test | average L2 Chamfer distance x1e4: 9.23 -> 7.60 | Improved |
Later papers that compared with this method
17 later accepted papers report SnowflakeNet as a baseline in their results tables; 17 of them report a better number on most of the results they share with it.
- SplAttN: Bridging 2D and 3D with Gaussian Soft Splatting and Attention for Point Cloud Completion (ICML 2026 · better on 3 of 3 shared results)
- PQDT: Pseudo-Query Dual Transformer for Robust Point Cloud Restoration (CVPR 2026 · better on 1 of 1 shared result)
- Semantic Guided Part Relation-aware Network for Point Cloud Completion (AAAI 2026 · better on 1 of 1 shared result)
- PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · better on 1 of 1 shared result)
- GenPC: Zero-shot Point Cloud Completion via 3D Generative Priors (CVPR 2025 · better on 1 of 1 shared result)
- PointSea: Point Cloud Completion via Self-structure Augmentation (IJCV 2025 · better on 1 of 1 shared result)
- PointAttN: You Only Need Attention for Point Cloud Completion (AAAI 2024 · better on 2 of 2 shared results)
- ProtoComp: Diverse Point Cloud Completion with Controllable Prototype (ECCV 2024 · better on 2 of 2 shared results)
9 more not listed.
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- PMP-Net — PMP-Net: Point Cloud Completion by Learning Multi-Step Point Moving Paths (CVPR 2021) · better on 3 of 3 shared results
- GRNet — GRNet: Gridding Residual Network for Dense Point Cloud Completion (ECCV 2020) · better on 2 of 2 shared results
- FoldingNet — FoldingNet: Point Cloud Auto-Encoder via Deep Grid Deformation (CVPR 2018) · better on 2 of 2 shared results
- SoftPoolNet — SoftPoolNet: Shape Descriptor for Point Cloud Completion and Classification (ECCV 2020) · 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
- Snowflake Point Deconvolution for Point Cloud Completion and Generation with Skip-Transformer (TPAMI 2022 · similarity 0.98)
- Point Cloud Completion by Skip-Attention Network With Hierarchical Folding (CVPR 2020 · similarity 0.92)
- Morphing and Sampling Network for Dense Point Cloud Completion (AAAI 2020 · similarity 0.91)
- ASHF-Net: Adaptive Sampling and Hierarchical Folding Network for Robust Point Cloud Completion (AAAI 2021 · similarity 0.90)
- SymmCompletion: High-Fidelity and High-Consistency Point Cloud Completion with Symmetry Guidance (AAAI 2025 · similarity 0.89)
- CRA-PCN: Point Cloud Completion with Intra- and Inter-level Cross-Resolution Transformers (AAAI 2024 · similarity 0.89)
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.
