Zoom In, Zoom Out: Injecting Scale Invariance into Landuse Classification CNNs

Jesse Murray, Diego Marcos, Devis Tuia

Research output: Chapter in Book/Report/Conference proceedingConference paperAcademicpeer-review

Abstract

We propose a Convolutional Neural Network (CNN), which encodes local scale invariance and equivariance in a multiresolution, multi-sensor image classification task. We show that the locally scale invariant model achieves results that are in line with state-of-the-art. The scale invariant and equivariant models also prove to be more robust to reductions in training data and number of filters used in each convolutional layer. These results demonstrate the benefit of disentangling scale within the learned features of CNNs, in particular when processing multi-resolution imagery. This is beneficial in the two studied cases: when training data is limited, or when the number of model parameters must be kept to a minimum.
Original languageEnglish
Title of host publicationIGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium
Subtitle of host publicationProceedings
PublisherIEEE
Pages5240-5243
ISBN (Electronic)9781538691540
ISBN (Print)9781538691557
DOIs
Publication statusPublished - 14 Nov 2019
EventIGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium - Yokohama, Japan
Duration: 28 Jul 20192 Aug 2019

Conference

ConferenceIGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium
Period28/07/192/08/19

Fingerprint Dive into the research topics of 'Zoom In, Zoom Out: Injecting Scale Invariance into Landuse Classification CNNs'. Together they form a unique fingerprint.

Cite this