English

Projective Manifold Gradient Layer for Deep Rotation Regression

Computer Vision and Pattern Recognition 2022-03-31 v3

Abstract

Regressing rotations on SO(3) manifold using deep neural networks is an important yet unsolved problem. The gap between the Euclidean network output space and the non-Euclidean SO(3) manifold imposes a severe challenge for neural network learning in both forward and backward passes. While several works have proposed different regression-friendly rotation representations, very few works have been devoted to improving the gradient backpropagating in the backward pass. In this paper, we propose a manifold-aware gradient that directly backpropagates into deep network weights. Leveraging Riemannian optimization to construct a novel projective gradient, our proposed regularized projective manifold gradient (RPMG) method helps networks achieve new state-of-the-art performance in a variety of rotation estimation tasks. Our proposed gradient layer can also be applied to other smooth manifolds such as the unit sphere. Our project page is at https://jychen18.github.io/RPMG.

Keywords

Cite

@article{arxiv.2110.11657,
  title  = {Projective Manifold Gradient Layer for Deep Rotation Regression},
  author = {Jiayi Chen and Yingda Yin and Tolga Birdal and Baoquan Chen and Leonidas Guibas and He Wang},
  journal= {arXiv preprint arXiv:2110.11657},
  year   = {2022}
}

Comments

CVPR2022

R2 v1 2026-06-24T07:05:58.654Z