mlsanityData to 7 Oct 2026

3D Equivariant Pose Regression via Direct Wigner-D Harmonics Prediction

arXiv 2411.00543 ↗First posted 1 Nov 2024cs.CV cs.LG cs.ROcited by 4

Not on the map. Why?

Abstract (arXiv)

Determining the 3D orientations of an object in an image, known as single-image pose estimation, is a crucial task in 3D vision applications. Existing methods typically learn 3D rotations parametrized in the spatial domain using Euler angles or quaternions, but these representations often introduce discontinuities and singularities. SO(3)-equivariant networks enable the structured capture of pose patterns with data-efficient learning, but the parametrizations in spatial domain are incompatible with their architecture, particularly spherical CNNs, which operate in the frequency domain to enhance computational efficiency. To overcome these issues, we propose a frequency-domain approach that directly predicts Wigner-D coefficients for 3D rotation regression, aligning with the operations of spherical CNNs. Our SO(3)-equivariant pose harmonics predictor overcomes the limitations of spatial parameterizations, ensuring consistent pose estimation under arbitrary rotations. Trained with a frequency-domain regression loss, our method achieves state-of-the-art results on benchmarks such as ModelNet10-SO(3) and PASCAL3D+, with significant improvements in accuracy, robustness, and data efficiency.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Builds on 58 of its 82 references that are in the map. First posted on arXiv 1 Nov 2024.

Builds on

58 of 82 references.

References · 58

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  2. 2018On the Continuity of Rotation Representations in Neural Networkscited by 1,186
  3. 2016Group Equivariant Convolutional Networkscited by 1,154
  4. 2017PoseCNN: A Convolutional Neural Network for 6D Object Pose Estimation in Cluttered Scenescited by 768
  5. 2019Normalized Object Coordinate Space for Category-Level 6D Object Pose and Size Estimationcited by 490
  6. 2018Spherical CNNscited by 452
  7. 2019General E(2)-Equivariant Steerable CNNscited by 441
  8. 2017Learning SO(3) Equivariant Representations with Spherical CNNscited by 244
  9. 2017Geometric Loss Functions for Camera Pose Regression with Deep Learningcited by 224
  10. 2016DeMoN: Depth and Motion Network for Learning Monocular Stereocited by 200
  11. 2018Clebsch-Gordan Nets: a Fully Fourier Space Spherical Convolutional Neural Networkcited by 168
  12. 2022CroCo: Self-Supervised Pre-training for 3D Vision Tasks by Cross-View Completioncited by 158
38 more
  1. 2023PoseDiffusion: Solving Pose Estimation via Diffusion-aided Bundle Adjustmentcited by 146
  2. 2024Cameras as Rays: Pose Estimation via Ray Diffusioncited by 129
  3. 2022RelPose: Predicting Probabilistic Relative Rotation for Single Objects in the Wildcited by 102
  4. 2022LaMAR: Benchmarking Localization and Mapping for Augmented Realitycited by 98
  5. 2019Overcoming Limitations of Mixture Density Networks: A Sampling and Fitting Framework for Multimodal Future Predictioncited by 96
  6. 2015Render for CNN: Viewpoint Estimation in Images Using CNNs Trained with Rendered 3D Model Viewscited by 94
  7. 2021Volumetric Grasping Network: Real-time 6 DOF Grasp Detection in Cluttercited by 94
  8. 2023RelPose++: Recovering 6D Poses from Sparse-view Observationscited by 87
  9. 2020An Analysis of SVD for Deep Rotation Estimationcited by 85
  10. 2021Deep Regression on Manifolds: A 3D Rotation Case Studycited by 67
  11. 2018Deep Part Induction from Articulated Object Pairscited by 66
  12. 2020DeepSphere: a graph-based spherical CNNcited by 65
  13. 2021Implicit-PDF: Non-Parametric Representation of Probability Distributions on the Rotation Manifoldcited by 65
  14. 2014Viewpoints and Keypointscited by 59
  15. 2019Reconstructing continuous distributions of 3D protein structure from cryo-EM imagescited by 56
  16. 2020Spin-Weighted Spherical CNNscited by 50
  17. 2021Wide-Baseline Relative Camera Pose Estimation with Directional Learningcited by 49
  18. 2021MultiBodySync: Multi-Body Segmentation and Motion Estimation via 3D Scan Synchronizationcited by 46
  19. 2018Deep Directional Statistics: Pose Estimation with Uncertainty Quantificationcited by 45
  20. 2020Deep Bingham Networks: Dealing with Uncertainty and Ambiguity in Pose Estimationcited by 43
  21. 2021Extreme Rotation Estimation using Dense Correlation Volumescited by 42
  22. 2019Quaternion Equivariant Capsule Networks for 3D Point Cloudscited by 41
  23. 2024Learning with 3D rotations, a hitchhiker's guide to SO(3)cited by 39
  24. 2019Spherical Regression: Learning Viewpoints, Surface Normals and 3D Rotations on n-Spherescited by 37
  25. 2020Probabilistic orientation estimation with matrix Fisher distributionscited by 36
  26. 2021Projective Manifold Gradient Layer for Deep Rotation Regressioncited by 31
  27. 2020Efficient Generalized Spherical CNNscited by 29
  28. 2018Deep-6DPose: Recovering 6D Object Pose from a Single RGB Imagecited by 26
  29. 2022Self-Supervised Equivariant Learning for Oriented Keypoint Detectioncited by 26
  30. 20206D Camera Relocalization in Ambiguous Scenes via Continuous Multimodal Inferencecited by 23
  31. 2023Learning Rotation-Equivariant Features for Visual Correspondencecited by 23
  32. 2023Scaling Spherical CNNscited by 21
  33. 2022Unified Fourier-based Kernel and Nonlinearity Design for Equivariant Networks on Homogeneous Spacescited by 20
  34. 2018A Mixed Classification-Regression Framework for 3D Pose Estimation from 2D Imagescited by 19
  35. 2023Image to Sphere: Learning Equivariant Features for Efficient Pose Predictioncited by 17
  36. 2022FisherMatch: Semi-Supervised Rotation Regression via Entropy-based Filteringcited by 16
  37. 2023Delving into Discrete Normalizing Flows on SO(3) Manifold for Probabilistic Rotation Modelingcited by 14
  38. 2018Occlusion Resistant Object Rotation Regression from Point Cloud Segmentscited by 9

50 most cited of 58 All 58 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 3; 2026: 1.

4 papers.

  1. 2025Learning Shape-Independent Transformation via Spherical Representations for Category-Level Object Pose Estimationcited by 8
  2. 2025Axis-level Symmetry Detection with Group-Equivariant Representationcited by 2
  3. 2025Equivariant symmetry-aware head pose estimation for fetal MRIcited by 0
  4. 2026REViT: Roto-reflection Equivariant Convolutional Vision Transformercited by 0