On translation invariance in CNNs: Convolutional layers can exploit absolute spatial location

Research output: Chapter in Book/Conference proceedings/Edited volumeConference contributionScientificpeer-review

4 Citations (Scopus)

Abstract

In this paper we challenge the common assumption that convolutional layers in modern CNNs are translation invariant. We show that CNNs can and will exploit the absolute spatial location by learning filters that respond exclusively to particular absolute locations by exploiting image boundary effects. Because modern CNNs filters have a huge receptive field, these boundary effects operate even far from the image boundary, allowing the network to exploit absolute spatial location all over the image. We give a simple solution to remove spatial location encoding which improves translation invariance and thus gives a stronger visual inductive bias which particularly benefits small data sets. We broadly demonstrate these benefits on several architectures and various applications such as image classification, patch matching, and two video classification datasets.

Original languageEnglish
Title of host publication2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Subtitle of host publicationProceedings
EditorsL. O'Conner
Place of PublicationPiscataway
PublisherIEEE
Pages14262-14273
Number of pages12
ISBN (Electronic)978-1-7281-7168-5
ISBN (Print)978-1-7281-7169-2
DOIs
Publication statusPublished - 2020
Event2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2020 - Virtual, Online, United States
Duration: 14 Jun 202019 Jun 2020

Conference

Conference2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2020
CountryUnited States
CityVirtual, Online
Period14/06/2019/06/20

Fingerprint Dive into the research topics of 'On translation invariance in CNNs: Convolutional layers can exploit absolute spatial location'. Together they form a unique fingerprint.

Cite this