mlsanityData to 14 Oct 2026

Multi-body SE(3) Equivariance for Unsupervised Rigid Segmentation and Motion Estimation

arXiv 2306.05584 ↗First posted 8 Jun 2023cs.CV cs.AI cs.LGcited by 7

Not on the map. Why?

Abstract (arXiv)

A truly generalizable approach to rigid segmentation and motion estimation is fundamental to 3D understanding of articulated objects and moving scenes. In view of the closely intertwined relationship between segmentation and motion estimates, we present an SE(3) equivariant architecture and a training strategy to tackle this task in an unsupervised manner. Our architecture is composed of two interconnected, lightweight heads. These heads predict segmentation masks using point-level invariant features and estimate motion from SE(3) equivariant features, all without the need for category information. Our training strategy is unified and can be implemented online, which jointly optimizes the predicted segmentation and motion by leveraging the interrelationships among scene flow, segmentation mask, and rigid transformations. We conduct experiments on four datasets to demonstrate the superiority of our method. The results show that our method excels in both model performance and computational efficiency, with only 0.25M parameters and 0.92G FLOPs. To the best of our knowledge, this is the first work designed for category-agnostic part-level SE(3) equivariance in dynamic point clouds.

Key facts. Cited by 7 papers in the map (data through 14 Oct 2026). Builds on 52 of its 80 references that are in the map. First posted on arXiv 8 Jun 2023.

Builds on

52 of 80 references.

References · 52

  1. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,523
  2. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
  3. 2017ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenescited by 3,117
  4. 2015ShapeNet: An Information-Rich 3D Model Repositorycited by 2,897
  5. 2018PointPillars: Fast Encoders for Object Detection from Point Cloudscited by 1,715
  6. 2019KPConv: Flexible and Deformable Convolution for Point Cloudscited by 1,205
  7. 2016Group Equivariant Convolutional Networkscited by 1,165
  8. 2020Center-based 3D Object Detection and Trackingcited by 1,144
  9. 2018PointRCNN: 3D Object Proposal Generation and Detection from Point Cloudcited by 941
  10. 2020Convolutional Occupancy Networkscited by 761
  11. 2017Frustum PointNets for 3D Object Detection from RGB-D Datacited by 640
  12. 2020SAPIEN: A SimulAted Part-based Interactive ENvironmentcited by 595
38 more
  1. 20163D-R2N2: A Unified Approach for Single and Multi-view 3D Object Reconstructioncited by 588
  2. 2019RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Cloudscited by 585
  3. 2018Spherical CNNscited by 456
  4. 2016Steerable CNNscited by 305
  5. 2021Vector Neurons: A General Framework for SO(3)-Equivariant Networkscited by 305
  6. 2019Gauge Equivariant Convolutional Networks and the Icosahedral CNNcited by 267
  7. 2021Geometric and Physical Quantities Improve E(3) Equivariant Message Passingcited by 228
  8. 2017Learning Steerable Filters for Rotation Equivariant CNNscited by 194
  9. 2021Frame Averaging for Invariant and Equivariant Network Designcited by 168
  10. 2019Learning Object Bounding Boxes for 3D Instance Segmentation on Point Cloudscited by 156
  11. 2019MeteorNet: Deep Learning on Dynamic 3D Point Cloud Sequencescited by 117
  12. 2020P2B: Point-to-Box Network for 3D Object Tracking in Point Cloudscited by 84
  13. 2021Equivariant Point Network for 3D Point Cloud Analysiscited by 81
  14. 2020Towards Semantic Segmentation of Urban-Scale 3D Point Clouds: A Dataset, Benchmarks and Challengescited by 80
  15. 2020Learning multiview 3D point cloud registrationcited by 78
  16. 2019Leveraging Shape Completion for 3D Siamese Trackingcited by 74
  17. 2020FlowStep3D: Model Unrolling for Self-Supervised Scene Flow Estimationcited by 71
  18. 2021You Only Hypothesize Once: Point Cloud Registration with Rotation-equivariant Descriptorscited by 68
  19. 2022PSTNet: Point Spatio-Temporal Convolution on Point Cloud Sequencescited by 68
  20. 2018Deep Part Induction from Articulated Object Pairscited by 66
  21. 2022Learning Hierarchical Protein Representations via Complete 3D Graph Networkscited by 55
  22. 2020CaSPR: Learning Canonical Spatiotemporal Point Cloud Representationscited by 51
  23. 2022Dynamic 3D Scene Analysis by Point Cloud Accumulationcited by 47
  24. 2021MultiBodySync: Multi-Body Segmentation and Motion Estimation via 3D Scan Synchronizationcited by 46
  25. 2021Leveraging SE(3) Equivariance for Self-Supervised Category-Level Object Pose Estimationcited by 45
  26. 2022Bi-PointFlowNet: Bidirectional Learning for Point Cloud Based Scene Flow Estimationcited by 38
  27. 2020CONSAC: Robust Multi-Model Fitting by Conditional Sample Consensuscited by 30
  28. 2022SE(3)-Equivariant Attention Networks for Shape Reconstruction in Function Spacecited by 27
  29. 2022No Pain, Big Gain: Classify Dynamic Point Cloud Sequences with Static Models by Fitting Feature-level Space-time Surfacescited by 25
  30. 2022OGC: Unsupervised 3D Object Segmentation from Rigid Dynamics of Point Cloudscited by 25
  31. 2021Learning Compositional Representation for 4D Captures with Neural ODEcited by 22
  32. 2022Rotationally Equivariant 3D Object Detectioncited by 21
  33. 2020Self-Supervised Learning of Part Mobility from Point Cloud Sequencecited by 18
  34. 2023EFEM: Equivariant Neural Field Expectation Maximization for 3D Object Segmentation Without Scene Supervisioncited by 18
  35. 2023Generalizing Neural Human Fitting to Unseen Poses With Articulated SE(3) Equivariancecited by 14
  36. 2022M"obius Convolutions for Spherical CNNscited by 13
  37. 20193D Rigid Motion Segmentation with Mixed and Unknown Number of Modelscited by 11
  38. 2020ASAP-Net: Attention and Structure Aware Point Cloud Sequence Segmentationcited by 8

50 most cited of 52 All 52 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 2; 2026: 5.

7 papers.

  1. 2025ETCH: Generalizing Body Fitting to Clothed Humans via Equivariant Tightnesscited by 9
  2. 2026Articulation in Motion: Prior-free Part Mobility Analysis for Articulated Objects By Dynamic-Static Disentanglementcited by 6
  3. 2025sim2art: Accurate Articulated Object Modeling from a Single Video using Synthetic Training Data Onlycited by 2
  4. 2026RINO: Rotation-Invariant Non-Rigid Correspondencescited by 2
  5. 2026Articulation in Prime: Primitive-Based Articulated Object Understanding from a Single Casual Videocited by 1
  6. 2026MotionBits: Video Segmentation through Motion-Level Analysis of Rigid Bodiescited by 0
  7. 2026GAUGE: Group-Wise View-Inconsistency Rectification for Feed-Forward 4D Trackingcited by 0