Header

UZH-Logo

Maintenance Infos

Rotation equivariant vector field networks


Marcos, Diego; Volpi, Michele; Komodakis, Nikos; Tuia, Devis (2017). Rotation equivariant vector field networks. In: 2017 IEEE International Conference on Computer Vision (ICCV), Venedig, 22 October 2017 - 29 October 2017, 5058-5067.

Abstract

In many computer vision tasks, we expect a particular behavior of the output with respect to rotations of the input image. If this relationship is explicitly encoded, instead of treated as any other variation, the complexity of the problem is decreased, leading to a reduction in the size of the required model. In this paper, we propose the Rotation Equivariant Vector Field Networks (RotEqNet), a Convolutional Neural Network (CNN) architecture encoding rotation equivariance, invariance and covariance. Each convolutional filter is applied at multiple orientations and returns a vector field representing magnitude and angle of the highest scoring orientation at every spatial location. We develop a modified convolution operator relying on this representation to obtain deep architectures. We test RotEqNet on several problems requiring different responses with respect to the inputs' rotation: image classification, biomedical image segmentation, orientation estimation and patch matching. In all cases, we show that RotEqNet offers extremely compact models in terms of number of parameters and provides results in line to those of networks orders of magnitude larger.

Abstract

In many computer vision tasks, we expect a particular behavior of the output with respect to rotations of the input image. If this relationship is explicitly encoded, instead of treated as any other variation, the complexity of the problem is decreased, leading to a reduction in the size of the required model. In this paper, we propose the Rotation Equivariant Vector Field Networks (RotEqNet), a Convolutional Neural Network (CNN) architecture encoding rotation equivariance, invariance and covariance. Each convolutional filter is applied at multiple orientations and returns a vector field representing magnitude and angle of the highest scoring orientation at every spatial location. We develop a modified convolution operator relying on this representation to obtain deep architectures. We test RotEqNet on several problems requiring different responses with respect to the inputs' rotation: image classification, biomedical image segmentation, orientation estimation and patch matching. In all cases, we show that RotEqNet offers extremely compact models in terms of number of parameters and provides results in line to those of networks orders of magnitude larger.

Statistics

Citations

Dimensions.ai Metrics

Altmetrics

Downloads

1 download since deposited on 23 Mar 2018
1 download since 12 months
Detailed statistics

Additional indexing

Item Type:Conference or Workshop Item (Paper), not_refereed, original work
Communities & Collections:07 Faculty of Science > Institute of Geography
Dewey Decimal Classification:910 Geography & travel
Language:English
Event End Date:29 October 2017
Deposited On:23 Mar 2018 15:03
Last Modified:31 Jul 2018 06:03
Publisher:IEEE
ISBN:978-1-5386-1032-9
OA Status:Closed
Publisher DOI:https://doi.org/10.1109/ICCV.2017.540

Download

Content: Published Version
Language: English
Filetype: PDF - Registered users only
Size: 997kB
View at publisher