M\"obius Transform for Mitigating Perspective Distortions in Representation Learning

arXiv:2405.02296v1 Announce Type: new
Abstract: Perspective distortion (PD) causes unprecedented changes in shape, size, orientation, angles, and other spatial relationships of visual concepts in images. Precisely estimating camera intrinsic and extrinsic parameters is a challenging task that prevents synthesizing perspective distortion. Non-availability of dedicated training data poses a critical barrier to developing robust computer vision methods. Additionally, distortion correction methods make other computer vision tasks a multi-step approach and lack performance. In this work, we propose mitigating perspective distortion (MPD) …

abstract arxiv availability concepts cs.cv data images intrinsic parameters perspective relationships representation representation learning spatial training training data type visual visual concepts