3D-Rotation-Equivariant Quaternion Neural Networks
This addresses the need for more robust 3D vision models for applications like robotics or autonomous driving, though it appears incremental as it modifies existing networks rather than introducing a new paradigm.
The paper tackles the problem of achieving rotation-equivariance in 3D point cloud processing by proposing rules to revise neural networks into rotation-equivariant quaternion neural networks (REQNNs), resulting in higher rotation robustness compared to original networks.
This paper proposes a set of rules to revise various neural networks for 3D point cloud processing to rotation-equivariant quaternion neural networks (REQNNs). We find that when a neural network uses quaternion features under certain conditions, the network feature naturally has the rotation-equivariance property. Rotation equivariance means that applying a specific rotation transformation to the input point cloud is equivalent to applying the same rotation transformation to all intermediate-layer quaternion features. Besides, the REQNN also ensures that the intermediate-layer features are invariant to the permutation of input points. Compared with the original neural network, the REQNN exhibits higher rotation robustness.