Research map / Point clouds / Shape and surface completion
Structure-guided completion
Complete shapes using skip-attention, structure priors, diffusion refinement and CLIP guidance.
22 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2019–2026) · an approach in Shape and surface completion, keyword cluster “cloud completion · complete · shape completion”
State History Benchmarks Tried Checklist Preprints Export
Where it stands
Still moving The best reported result is still improving on 1 of 1 tracked benchmark.
9 papers in 2023–24 and 7 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 10.1% to 6.2%.
Open questions the data points to
- KITTI (MMD): best went from 0.458 (up to 2024) to 0.353 in 2025 — PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · 7 citations)
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2019 · TopNet: Structural Point Cloud Decoder (CVPR 2019 · 532 citations)
- 2020 · Point Cloud Completion by Skip-Attention Network With Hierarchical Folding (CVPR 2020 · 288 citations)
- 2021 · PMP-Net: Point Cloud Completion by Learning Multi-Step Point Moving Paths (CVPR 2021 · 206 citations)
- 2022 · A Conditional Point Diffusion-Refinement Paradigm for 3D Point Cloud Completion (ICLR 2022 · 8 citations)
- 2023 · ProxyFormer: Proxy Alignment Assisted Point Cloud Completion With Missing Part Sensitive Transformer (CVPR 2023 · 93 citations)
- 2024 · PointAttN: You Only Need Attention for Point Cloud Completion (AAAI 2024 · 78 citations)
- 2025 · SuperPC: A Single Diffusion Model for Point Cloud Completion, Upsampling, Denoising, and Colorization (CVPR 2025 · 9 citations)
- 2026 · Semantic Guided Part Relation-aware Network for Point Cloud Completion (AAAI 2026 · 1 citation)
Earliest papers here
- TopNet: Structural Point Cloud Decoder (CVPR 2019 · 532 citations)
- Point Cloud Completion by Skip-Attention Network With Hierarchical Folding (CVPR 2020 · 288 citations)
- Unpaired Point Cloud Completion on Real Scans using Adversarial Training (ICLR 2020 · 83 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
KITTI · MMD (lower is better)
5 papers, 2023–2025 · still improving
| Year | Best so far | Set by |
|---|---|---|
| 2023 | 0.458 | AnchorFormer: Point Cloud Completion From Discriminative Nodes (CVPR 2023 · 83 citations) |
| 2025 | 0.353 | PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · 7 citations) |
What has been tried
20 papers here report 234 measured changes (a component added, removed or swapped, compared with a baseline): helped 74%, no clear effect 5%, hurt 18%, mixed 3%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| add GDP units to the feature extractor | Completion3D · CD-L2 (Avg) | 7.42 -> 6.63 | Helped | PointAttN: You Only Need Attention for Point Cloud Completion (AAAI 2024 · 78 citations) |
| Replace FoldingNet seed generation with ODGNet Seed Generation U-Net | PCN test · CD-L1 (×10−3) | 14.31 -> 7.59 | Helped | Orthogonal Dictionary Guided Shape Completion Network for Point Cloud (AAAI 2024 · 31 citations) |
| add refinement network | SemanticKITTI val · CD [m] | 0.434 -> 0.375 | Helped | Scaling Diffusion Models to Real-World 3D LiDAR Scene Completion (CVPR 2024 · 28 citations) |
| add inter-level and intra-level CRT | PCN · CD-ℓ1 | 7.85 -> 6.39 | Helped | CRA-PCN: Point Cloud Completion with Intra- and Inter-level Cross-Resolution Transformers (AAAI 2024 · 25 citations) |
| Add phase I dual-phase structure | Module ablation · Chamfer Distance | 1.850 -> 1.465 | Helped | CDPNet: Cross-Modal Dual Phases Network for Point Cloud Completion (AAAI 2024 · 17 citations) |
| Add LGRP | PCN · CD-ℓ1 | 6.92 -> 6.47 | Helped | PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · 7 citations) |
| add encoder-codebook (shallow-level quantization) | PCN · CD-l1 (x10^3) | -0.05 | Helped | DC-PCN: Point Cloud Completion Network with Dual-Codebook Guided Quantization (AAAI 2025 · 5 citations) |
| Add the position-aware module to the CLIP-enhanced model | PCN · L1 CD × 10^3 | 7.19 -> 7.17 | Helped | Position-Aware Guided Point Cloud Completion with CLIP Model (AAAI 2025 · 3 citations) |
| add VI-PointConv and input normals | ShapeNet chair completion, arbitrary pose and scale · One-sided CD | 7.12 -> 6.04 | Helped | Point-based Instance Completion with Scene Constraints (ICLR 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 6 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- PCN · 3 of 6 papers
- KITTI · 3 of 6 papers
- ShapeNet34 · 2 of 6 papers
- ShapeNet · 2 of 6 papers
Baselines
- AdaPoinTr · 4 of 6 papers
- SeedFormer · 4 of 6 papers
- FoldingNet · 3 of 6 papers
- PCN · 3 of 6 papers
- GRNet · 3 of 6 papers
- PoinTr · 3 of 6 papers
- TopNet · 2 of 6 papers
- CRN · 2 of 6 papers
Recent preprints
Loading recent arXiv preprints…
Use it
Download BibTeX of 22 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Point cloud completion networks (50 papers)
Elsewhere in the topic
- Unsupervised shape correspondence (102 papers)
- Point cloud upsampling and normals (64 papers)
- Distance-function surface reconstruction (50 papers)
- Point cloud diffusion generation (44 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).