Geometry Aware Convolutional Filters for Omnidirectional Images Representation

Renata Khasanova, Pascal Frossard
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:3351-3359, 2019.

Abstract

Due to their wide field of view, omnidirectional cameras are frequently used by autonomous vehicles, drones and robots for navigation and other computer vision tasks. The images captured by such cameras, are often analyzed and classified with techniques designed for planar images that unfortunately fail to properly handle the native geometry of such images and therefore results in suboptimal performance. In this paper we aim at improving popular deep convolutional neural networks so that they can properly take into account the specific properties of omnidirectional data. In particular we propose an algorithm that adapts convolutional layers, which often serve as a core building block of a CNN, to the properties of omnidirectional images. Thus, our filters have a shape and size that adapt to the location on the omnidirectional image. We show that our method is not limited to spherical surfaces and is able to incorporate the knowledge about any kind of projective geometry inside the deep learning network. As depicted by our experiments, our method outperforms the existing deep neural network techniques for omnidirectional image classification and compression tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-khasanova19a, title = {Geometry Aware Convolutional Filters for Omnidirectional Images Representation}, author = {Khasanova, Renata and Frossard, Pascal}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {3351--3359}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/khasanova19a/khasanova19a.pdf}, url = {https://proceedings.mlr.press/v97/khasanova19a.html}, abstract = {Due to their wide field of view, omnidirectional cameras are frequently used by autonomous vehicles, drones and robots for navigation and other computer vision tasks. The images captured by such cameras, are often analyzed and classified with techniques designed for planar images that unfortunately fail to properly handle the native geometry of such images and therefore results in suboptimal performance. In this paper we aim at improving popular deep convolutional neural networks so that they can properly take into account the specific properties of omnidirectional data. In particular we propose an algorithm that adapts convolutional layers, which often serve as a core building block of a CNN, to the properties of omnidirectional images. Thus, our filters have a shape and size that adapt to the location on the omnidirectional image. We show that our method is not limited to spherical surfaces and is able to incorporate the knowledge about any kind of projective geometry inside the deep learning network. As depicted by our experiments, our method outperforms the existing deep neural network techniques for omnidirectional image classification and compression tasks.} }
Endnote
%0 Conference Paper %T Geometry Aware Convolutional Filters for Omnidirectional Images Representation %A Renata Khasanova %A Pascal Frossard %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-khasanova19a %I PMLR %P 3351--3359 %U https://proceedings.mlr.press/v97/khasanova19a.html %V 97 %X Due to their wide field of view, omnidirectional cameras are frequently used by autonomous vehicles, drones and robots for navigation and other computer vision tasks. The images captured by such cameras, are often analyzed and classified with techniques designed for planar images that unfortunately fail to properly handle the native geometry of such images and therefore results in suboptimal performance. In this paper we aim at improving popular deep convolutional neural networks so that they can properly take into account the specific properties of omnidirectional data. In particular we propose an algorithm that adapts convolutional layers, which often serve as a core building block of a CNN, to the properties of omnidirectional images. Thus, our filters have a shape and size that adapt to the location on the omnidirectional image. We show that our method is not limited to spherical surfaces and is able to incorporate the knowledge about any kind of projective geometry inside the deep learning network. As depicted by our experiments, our method outperforms the existing deep neural network techniques for omnidirectional image classification and compression tasks.
APA
Khasanova, R. & Frossard, P.. (2019). Geometry Aware Convolutional Filters for Omnidirectional Images Representation. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:3351-3359 Available from https://proceedings.mlr.press/v97/khasanova19a.html.

Related Material