Research map / Point clouds / Point cloud classification
Point cloud detection and tracking
Apply point cloud networks to tracking, completion and anomaly detection tasks.
56 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Point cloud classification, keyword cluster “motion · estimation · scene”
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.
24 papers in 2023–24 and 11 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 10.3% to 4.2%.
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
- 2016 · Progressive Prioritized Multi-View Stereo (CVPR 2016 · 42 citations)
- 2017 · Quad-Networks: Unsupervised Learning to Rank for Interest Point Detection (CVPR 2017 · 186 citations)
- 2018 · Document Enhancement Using Visibility Detection (CVPR 2018 · 65 citations)
- 2019 · Revealing Scenes by Inverting Structure From Motion Reconstructions (CVPR 2019 · 118 citations)
- 2020 · Fusion-Aware Point Convolution for Online Semantic 3D Scene Segmentation (CVPR 2020 · 66 citations)
- 2021 · Three-dimensional spike localization and improved motion correction for Neuropixels recordings (NeurIPS 2021 · 40 citations)
- 2022 · A Keypoint-Based Global Association Network for Lane Detection (CVPR 2022 · 172 citations)
- 2023 · Multimodal Industrial Anomaly Detection via Hybrid Fusion (CVPR 2023 · 211 citations)
- 2024 · RCooper: A Real-world Large-scale Dataset for Roadside Cooperative Perception (CVPR 2024 · 58 citations)
- 2025 · Real-IAD D3: A Real-World 2D/Pseudo-3D/3D Dataset for Industrial Anomaly Detection (CVPR 2025 · 14 citations)
- 2026 · Towards High-Resolution 3D Anomaly Detection: A Scalable Dataset and Real-Time Framework for Subtle Industrial Defects (AAAI 2026 · 10 citations)
Earliest papers here
- Progressive Prioritized Multi-View Stereo (CVPR 2016 · 42 citations)
- Detecting Vanishing Points Using Global Image Context in a Non-Manhattan World (CVPR 2016 · 23 citations)
- Quad-Networks: Unsupervised Learning to Rank for Interest Point Detection (CVPR 2017 · 186 citations)
What has been tried
52 papers here report 507 measured changes (a component added, removed or swapped, compared with a baseline): helped 72%, no clear effect 6%, hurt 19%, 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 |
|---|---|---|---|---|
| exclude point clouds from MEMS LiDARs | Intersection · detection performance | Helped | RCooper: A Real-world Large-scale Dataset for Roadside Cooperative Perception (CVPR 2024 · 58 citations) | |
| add one decomposed orthogonal sparse completion convolution instead of none | SemanticKITTI val · mIoU | 12.1 -> 13.1 | Helped | SparseOcc: Rethinking Sparse Latent Representation for Vision-Based Semantic Occupancy Prediction (CVPR 2024 · 53 citations) |
| add top-view depth | KITTI validation split · RMSE (mm) | 763.56 -> 755.14 | Helped | Tri-Perspective View Decomposition for Geometry-Aware Depth Completion (CVPR 2024 · 42 citations) |
| add LSTM to the hierarchy-only model | IJRR random split, shape translation · T+R | 3.04 -> 2.77 | Helped | A Simple and Effective Point-based Network for Event Camera 6-DOFs Pose Relocalization (CVPR 2024 · 17 citations) |
| Downsample point cloud 4x instead of using original resolution | Real-IAD D³: connector housing-female, Hole · I-AUROC | 0.516 -> 0.500 | Hurt | Real-IAD D3: A Real-World 2D/Pseudo-3D/3D Dataset for Industrial Anomaly Detection (CVPR 2025 · 14 citations) |
| add the offset-direction loss Ldir | Anomaly-ShapeNet (15 categories ending in 0) · object-level AUC-ROC | 50.3 -> 84.2 | Helped | PO3AD: Predicting Point Offsets toward Better 3D Point Cloud Anomaly Detection (CVPR 2025 · 10 citations) |
| Use 2 template frames rather than 1 | KITTI All · Success | 70.1 -> 70.0 | No clear effect | M3SOT: Multi-Frame, Multi-Field, Multi-Space 3D Single Object Tracking (AAAI 2024 · 10 citations) |
| Increase the point grid from N=21 × 21 to N=41 × 41 | VIGOR same-area validation, known orientation · Mean localization error (m) | 2.50 -> 2.17 | Helped | FG^2: Fine-Grained Cross-View Localization by Fine-Grained Feature Matching (CVPR 2025 · 9 citations) |
| Initialize 3DGS with the MASt3R dense point cloud and poses instead of sparse SfM points and poses | Mip-NeRF 360 · PSNR | 16.6 -> 15.9 | Hurt | SPARS3R: Semantic Prior Alignment and Regularization for Sparse 3D Reconstruction (CVPR 2025 · 9 citations) |
| Add voxel-based ray sampling | Energy Landscape · geometric error (m) | 10.29 -> 10.05 | Helped | Efficient Privacy-Preserving Visual Localization Using 3D Ray Clouds (CVPR 2024 · 8 citations) |
| Use SIFT scale and orientation rather than random scale and orientation | HPatches viewpoint · Corner-warp AUC@5px | 13.5 -> 63.5 | Helped | Noisy One-point Homographies are Surprisingly Good (CVPR 2024 · 8 citations) |
| Add neural amplitude representation instead of pixel-wise amplitude representation | 3µm depth range · CRB (nm) | 65.5 -> 50.5 | Helped | CodedEvents: Optimal Point-Spread-Function Engineering for 3D-Tracking with Event Cameras (CVPR 2024 · 7 citations) |
What reviewers will expect
Datasets and baselines named in the 8 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Baselines
- M3DM · 2 of 8 papers
- PatchCore · 2 of 8 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 35 papers; from OpenAlex and the papers' affiliations).
- ShanghaiTech University · 3 papers
- Chinese Academy of Sciences · 2 papers
- Tsinghua University · 2 papers
- Tencent · 2 papers
- Shanghai Jiao Tong University · 2 papers
- Hanyang University · 2 papers
Use it
Download BibTeX of 37 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
- Scene flow and dynamic point clouds (89 papers)
- LiDAR point cloud compression and completion (60 papers)
Elsewhere in the topic
- Local feature aggregation networks (158 papers)
- Point-language and open-vocabulary 3D (107 papers)
- Equivariant point cloud networks (84 papers)
- Self-supervised point cloud pretraining (73 papers)
- Efficient point sampling architectures (45 papers)
- Adversarial point cloud attacks (37 papers)
- Masked point autoencoders (22 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).