mlsanityData to 14 Oct 2026

Learning general and distinctive 3D local deep descriptors for point cloud registration

arXiv 2105.10382 ↗First posted 21 May 2021cs.CV cs.ROcited by 42

Not on the map. Why?

Abstract (arXiv)

An effective 3D descriptor should be invariant to different geometric transformations, such as scale and rotation, robust to occlusions and clutter, and capable of generalising to different application domains. We present a simple yet effective method to learn general and distinctive 3D local descriptors that can be used to register point clouds that are captured in different domains. Point cloud patches are extracted, canonicalised with respect to their local reference frame, and encoded into scale and rotation-invariant compact descriptors by a deep neural network that is invariant to permutations of the input points. This design is what enables our descriptors to generalise across domains. We evaluate and compare our descriptors with alternative handcrafted and deep learning-based descriptors on several indoor and outdoor datasets that are reconstructed by using both RGBD sensors and laser scanners. Our descriptors outperform most recent descriptors by a large margin in terms of generalisation, and also become the state of the art in benchmarks where training and testing are performed in the same domain.

Key facts. Cited by 42 papers in the map (data through 14 Oct 2026). Builds on 18 of its 35 references that are in the map. First posted on arXiv 21 May 2021.

Builds on

18 of 35 references.

References · 18

  1. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,248
  2. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
  3. 2019KPConv: Flexible and Deformable Convolution for Point Cloudscited by 1,205
  4. 2018Open3D: A Modern Library for 3D Data Processingcited by 721
  5. 2018PPFNet: Global Context Aware Local Features for Robust 3D Point Matchingcited by 181
  6. 2020D3Feat: Joint Learning of Dense Detection and Description of 3D Local Featurescited by 177
  7. 2018The Perfect Match: 3D Point Cloud Matching with Smoothed Densitiescited by 169
  8. 2020SpinNet: Learning a General Surface Descriptor for 3D Point Cloud Registrationcited by 142
  9. 2018PPF-FoldNet: Unsupervised Learning of Rotation Invariant 3D Local Descriptorscited by 141
  10. 20183D Point Capsule Networkscited by 106
  11. 20183DFeat-Net: Weakly Supervised Local 3D Features for Point Cloud Registrationcited by 103
  12. 2020End-to-End Learning Local Multi-view Descriptors for 3D Point Cloudscited by 47
6 more
  1. 2017Learning Compact Geometric Featurescited by 45
  2. 2013Rotational Projection Statistics for 3D Local Surface Description and Object Recognitioncited by 36
  3. 20193D Local Features for Direct Pairwise Registrationcited by 26
  4. 2020Distinctive 3D local deep descriptorscited by 21
  5. 2019Learning an Effective Equivariant 3D Descriptor Without Supervisioncited by 14
  6. 2018Iterative Transformer Network for 3D Point Cloudcited by 11

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 3; 2023: 7; 2024: 8; 2025: 13; 2026: 11.

42 papers.

Sort:

By area:

  1. 2023FreeZe: Training-free zero-shot 6D pose estimation with geometric and vision foundation modelscited by 57
  2. 2025Any6D: Model-free 6D Pose Estimation of Novel Objectscited by 49
  3. 2022GIPSO: Geometrically Informed Propagation for Online Adaptation in 3D LiDAR Segmentationcited by 35
  4. 2023Open-vocabulary object 6D pose estimationcited by 29
  5. 2024KISS-Matcher: Fast and Robust Point Cloud Registration Revisitedcited by 16
  6. 2025Accurate and efficient zero-shot 6D pose estimation with frozen foundation modelscited by 13
  7. 2025TurboReg: TurboClique for Robust and Efficient Point Cloud Registrationcited by 13
  8. 2024UNOPose: Unseen Object Pose Estimation with an Unposed RGB-D Reference Imagecited by 12
  9. 2023G3Reg: Pyramid Graph-based Global Registration using Gaussian Ellipsoid Modelcited by 11
  10. 2022Overlap-guided Gaussian Mixture Models for Point Cloud Registrationcited by 10
  11. 2025Register Any Point: Scaling 3D Point Cloud Registration by Flow Matchingcited by 9
  12. 2023COLA: COarse-LAbel multi-source LiDAR semantic segmentation for autonomous drivingcited by 8
30 more
  1. 2025BUFFER-X: Towards Zero-Shot Point Cloud Registration in Diverse Scenescited by 8
  2. 2023Revisiting Fully Convolutional Geometric Features for Object 6D Pose Estimationcited by 7
  3. 2022Data Augmentation-free Unsupervised Learning for 3D Point Cloud Understandingcited by 6
  4. 2024PointDifformer: Robust Point Cloud Registration With Neural Diffusion and Transformercited by 6
  5. 2024Weak-to-Strong 3D Object Detection with X-Ray Distillationcited by 5
  6. 2025Unlocking Generalization Power in LiDAR Point Cloud Registrationcited by 5
  7. 2023Novel class discovery meets foundation models for 3D semantic segmentationcited by 4
  8. 2024ML-SemReg: Boosting Point Cloud Registration with Multi-level Semantic Consistencycited by 4
  9. 2024PointRegGPT: Boosting 3D Point Cloud Registration using Generative Point-Cloud Pairs for Trainingcited by 4
  10. 2025Distilling 3D distinctive local descriptors for 6D pose estimationcited by 4
  11. 2023PatchMixer: Rethinking network design to boost generalization for 3D point cloud understandingcited by 3
  12. 2025Fully-Geometric Cross-Attention for Point Cloud Registrationcited by 3
  13. 2025Automated 3D-GS Registration and Fusion via Skeleton Alignment and Gaussian-Adaptive Featurescited by 3
  14. 20243D Registration in 30 Years: A Surveycited by 2
  15. 2025CHIP: A multi-sensor dataset for 6D pose estimation of chairs in industrial settingscited by 2
  16. 2026UniMesh: Unifying 3D Mesh Understanding and Generationcited by 2
  17. 2024QuadricsReg: Large-Scale Point Cloud Registration using Quadric Primitivescited by 1
  18. 2025Correspondence-Free Fast and Robust Spherical Point Pattern Registrationcited by 1
  19. 2026Toward Zero-Shot Point Cloud Registration Across Diverse Scales, Scenes, and Sensor Setupscited by 1
  20. 2026COG: Confidence-aware Optimal Geometric Correspondence for Unsupervised Single-reference Novel Object Pose Estimationcited by 1
  21. 2025LAHNet: Local Attentive Hashing Network for Point Cloud Registrationcited by 0
  22. 2025AlignPose: Generalizable 6D Pose Estimation via Multi-view Feature-metric Alignmentcited by 0
  23. 2026Learning Cross-View Semantic Priors for Single-Reference Unseen Object Pose Estimationcited by 0
  24. 2026xperception -- Making Robotic Grasping Easiercited by 0
  25. 2026RRTrack: Robust and Recoverable Object 6D Pose Tracking for Dynamic Scenescited by 0
  26. 2026Coarse-to-Fine Registration of Jawbone CT and Intraoral Scan Data Using GeDi and ICP with Pseudo-IOS Ground Truthcited by 0
  27. 2026Foundational feature fusion for conditional flow matching in 6D pose estimationcited by 0
  28. 2026Back to the Feature: Zero-Shot 6DoF Pose Estimation via Dense Local Featurescited by 0
  29. 2026ReLoc: Rethinking Scene Coordinate Regression Architecture for Robust Outdoor LiDAR-based Localizationcited by 0
  30. 2026NovaSplat: Model-Free 6D Object Pose Estimation via 3D Gaussian Splatting Onboarding and Refinementcited by 0