Rotation-Invariant Transformer for Point Cloud Matching

arXiv:2303.08231v3 Announce Type: replace
Abstract: The intrinsic rotation invariance lies at the core of matching point clouds with handcrafted descriptors. However, it is widely despised by recent deep matchers that obtain the rotation invariance extrinsically via data augmentation. As the finite number of augmented rotations can never span the continuous SO(3) space, these methods usually show instability when facing rotations that are rarely seen. To this end, we introduce RoITr, a Rotation-Invariant Transformer to cope with the pose variations in …

abstract arxiv augmentation cloud continuous core cs.cv data however intrinsic lies rotation space transformer type via