Research map / Point clouds / Shape and surface completion
Point cloud completion networks
Complete partial point clouds with refinement, folding and cycle-based unpaired learning.
50 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2018–2026) · an approach in Shape and surface completion, keyword cluster “cloud completion · complete · shape completion”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
11 papers in 2023–24 and 19 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 12.4% to 16.8%.
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
- 2018 · Learning 3D Shape Completion From Laser Scan Data With Weak Supervision (CVPR 2018 · 250 citations)
- 2019 · RL-GAN-Net: A Reinforcement Learning Agent Controlled GAN Network for Real-Time Point Cloud Shape Completion (CVPR 2019 · 209 citations)
- 2020 · PF-Net: Point Fractal Network for 3D Point Cloud Completion (CVPR 2020 · 510 citations)
- 2021 · Variational Relational Point Completion Network (CVPR 2021 · 237 citations)
- 2022 · ShapeFormer: Transformer-Based Shape Completion via Sparse Representation (CVPR 2022 · 134 citations)
- 2023 · Generalized Deep 3D Shape Prior via Part-Discretized Diffusion Process (CVPR 2023 · 28 citations)
- 2024 · FSC: Few-point Shape Completion (CVPR 2024 · 12 citations)
- 2025 · SymmCompletion: High-Fidelity and High-Consistency Point Cloud Completion with Symmetry Guidance (AAAI 2025 · 11 citations)
- 2026 · Rethinking Multimodal Point Cloud Completion: A Completion-by-Correction Perspective (AAAI 2026 · 2 citations)
Earliest papers here
- Learning 3D Shape Completion From Laser Scan Data With Weak Supervision (CVPR 2018 · 250 citations)
- RL-GAN-Net: A Reinforcement Learning Agent Controlled GAN Network for Real-Time Point Cloud Shape Completion (CVPR 2019 · 209 citations)
- Leveraging Shape Completion for 3D Siamese Tracking (CVPR 2019 · 190 citations)
What has been tried
42 papers here report 440 measured changes (a component added, removed or swapped, compared with a baseline): helped 79%, no clear effect 3%, hurt 17%, mixed 2%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add a salient branch to the extensive branch | ShapeNet seen categories, 64 points · CD-ℓ1 | 10.53 -> 9.78 | Helped | FSC: Few-point Shape Completion (CVPR 2024 · 12 citations) |
| Add both key-geometry features Fk and missing-part features Fm as SGFormer guidance | PCN · CD (L1 × 10³) | 6.51 -> 6.28 | Helped | SymmCompletion: High-Fidelity and High-Consistency Point Cloud Completion with Symmetry Guidance (AAAI 2025 · 11 citations) |
| Add naive shape completion | KITTI Car · success | 63.0 -> 65.2 | Helped | Robust 3D Tracking with Quality-Aware Shape Completion (AAAI 2024 · 10 citations) |
| Fuse partial point cloud features with multi-view depth-image features instead of using depth images alone | PCN test · CD-Avg (ℓ1 ×103) | 17.69 -> 6.49 | Helped | PCDreamer: Point Cloud Completion Through Multi-view Diffusion Priors (CVPR 2025 · 8 citations) |
| Add SDS refining | Redwood · CD | 1.98 -> 1.74 | Helped | GenPC: Zero-shot Point Cloud Completion via 3D Generative Priors (CVPR 2025 · 7 citations) |
| add mixed sampling | ShapeNet-ViPC (ablation slice) · L2 Chamfer Distance × 10^-3 (Avg.) | 2.96 -> 2.21 | Helped | Digging into Intrinsic Contextual Information for High-fidelity 3D Point Cloud Completion (AAAI 2025 · 3 citations) |
| Replace polar coordinates with Cartesian coordinates | ABC test, architecture ablation · CD | 2.11 -> 3.34 | Hurt | Parametric Point Cloud Completion for Polygonal Surface Reconstruction (CVPR 2025 · 3 citations) |
| Add patch-based self-supervision rather than applying image-level self-supervision directly | Unbalanced MCOP images · completion sharpness | Helped | Self-Supervised Large Scale Point Cloud Completion for Archaeological Site Restoration (CVPR 2025 · 3 citations) | |
| add the local module | Shape-generator training set · reconstruction loss | 0.336 -> 0.153 | Helped | FaceCom: Towards High-fidelity 3D Facial Shape Completion via Optimization and Inpainting Guidance (CVPR 2024 · 3 citations) |
| Add curvature-based Ball Query anchor refinement with radius 0.075 | PCN subset · CD-L1 (×1000) | 14.74 -> 13.58 | Helped | ESCAPE: Equivariant Shape Completion via Anchor Point Encoding (CVPR 2025 · 2 citations) |
| add Bi-Aux Units | PCN · CD-ℓ1 | 6.62 -> 6.42 | Helped | PointMAC: Meta-Learned Adaptation for Robust Test-Time Point Cloud Completion (NeurIPS 2025 · 2 citations) |
| Add explicit geometric guidance via projected point clouds | RealBench · PSNR | 14.78 -> 16.37 | Helped | GeoComplete: Geometry-Aware Diffusion for Reference-Driven Image Completion (NeurIPS 2025 · 1 citation) |
What reviewers will expect
Datasets and baselines named in the 11 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 · 5 of 11 papers
- KITTI · 4 of 11 papers
- MVP · 3 of 11 papers
- ShapeNet-ViPC · 2 of 11 papers
- ShapeNet-55 · 2 of 11 papers
- ShapeNet-34 · 2 of 11 papers
Baselines
- PoinTr · 8 of 11 papers
- PCN · 6 of 11 papers
- SeedFormer · 5 of 11 papers
- SnowflakeNet · 5 of 11 papers
- FoldingNet · 3 of 11 papers
- TopNet · 3 of 11 papers
- GRNet · 3 of 11 papers
- AnchorFormer · 3 of 11 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 30 papers; from OpenAlex and the papers' affiliations).
- Chinese Academy of Sciences · 2 papers
- Harbin Institute of Technology · 2 papers
- Peng Cheng Laboratory · 2 papers
- Huawei · 2 papers
- Georgia Institute of Technology · 2 papers
Use it
Download BibTeX of 40 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
- Structure-guided completion (22 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).