Warped Convolutional Neural Networks For Large Homography Transformation with $\mathfrak{sl}(3)$ Algebra
Primary Area: representation learning for computer vision, audio, language, and other modalities
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Keywords: SL(3), Lie algebra, Equivariance, Homography Learning, Group Equivariant Architecture
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
TL;DR: We propose a Warped Convolution Networks to effectively learn the homography on algebra space with group convolution.
Abstract: Homography has fundamental and elegant relationship with the special linear group and its embedding Lie algebra structure. However, the integration of homography and algebraic expressions in neural networks remains largely unexplored. In this paper, we propose Warped Convolution Neural Networks to effectively learn and represent the homography by $\mathfrak{sl}(3)$ algebra with group convolution. Specifically, six commutative subgroups within the $SL(3)$ group are composed to form a homography. For each subgroup, a warp function is proposed to bridge the Lie algebra structure to its corresponding parameters in homography. By taking advantage of the warped convolution, homography learning is formulated into several simple pseudo-translation regressions. Our proposed method enables to learn features that are invariant to significant homography transformations through exploration along the Lie topology. Moreover, it can be easily plugged into other popular CNN-based methods and empower them with homography representation capability. Through extensive experiments on benchmark datasets such as POT, S-COCO, and MNIST-Proj, we demonstrate the effectiveness of our approach in various tasks like planar object tracking, homography estimation and classification.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 501
Loading