<?xml version="1.0" encoding="ISO-8859-1"?>
<metadatalist>
	<metadata ReferenceType="Journal Article">
		<site>mtc-m21d.sid.inpe.br 808</site>
		<holdercode>{isadg {BR SPINPE} ibi 8JMKD3MGPCW/3DT298S}</holdercode>
		<identifier>8JMKD3MGP3W34T/45HFHL8</identifier>
		<repository>sid.inpe.br/mtc-m21d/2021/10.04.15.37</repository>
		<lastupdate>2021:10.04.15.37.15 urlib.net/www/2021/06.04.03.40 simone</lastupdate>
		<metadatarepository>sid.inpe.br/mtc-m21d/2021/10.04.15.37.15</metadatarepository>
		<metadatalastupdate>2022:04.04.04.50.16 sid.inpe.br/bibdigital@80/2006/04.07.15.50 administrator</metadatalastupdate>
		<doi>10.1016/j.isprsjprs.2021.08.026</doi>
		<issn>0924-2716</issn>
		<citationkey>SotoVegaCoFeOrAlHeRo:2021:UnDoAd</citationkey>
		<title>An unsupervised domain adaptation approach for change detection and its application to deforestation mapping in tropical biomes</title>
		<year>2021</year>
		<month>Nov.</month>
		<typeofwork>journal article</typeofwork>
		<secondarytype>PRE PI</secondarytype>
		<numberoffiles>1</numberoffiles>
		<size>22364 KiB</size>
		<author>Soto Vega, Pedro Juan,</author>
		<author>Costa, Gilson Alexandre Ostwald Pedro da,</author>
		<author>Feitosa, Raul Queiroz,</author>
		<author>Ortega Adarme, Mabel Ximena,</author>
		<author>Almeida, Cláudio Aparecido de,</author>
		<author>Heipke, Christian,</author>
		<author>Rottensteiner, Franz,</author>
		<group></group>
		<group></group>
		<group></group>
		<group></group>
		<group>DIPE1-COGPI-INPE-MCTI-GOV-BR</group>
		<affiliation>Universidade Federal do Rio de Janeiro (UFRJ)</affiliation>
		<affiliation>Universidade do Estado do Rio de Janeiro (UERJ)</affiliation>
		<affiliation>Universidade Federal do Rio de Janeiro (UFRJ)</affiliation>
		<affiliation>Universidade Federal do Rio de Janeiro (UFRJ)</affiliation>
		<affiliation>Instituto Nacional de Pesquisas Espaciais (INPE)</affiliation>
		<affiliation>Leibniz Universitat Hannover (LUH)</affiliation>
		<affiliation>Leibniz Universitat Hannover (LUH)</affiliation>
		<electronicmailaddress></electronicmailaddress>
		<electronicmailaddress></electronicmailaddress>
		<electronicmailaddress></electronicmailaddress>
		<electronicmailaddress></electronicmailaddress>
		<electronicmailaddress>claudio.almeida@inpe.br</electronicmailaddress>
		<journal>ISPRS Journal of Photogrammetry and Remote Sensing</journal>
		<volume>181</volume>
		<pages>113-128</pages>
		<secondarymark>A1_GEOCIÊNCIAS A2_INTERDISCIPLINAR A2_CIÊNCIAS_AMBIENTAIS B1_ENGENHARIAS_IV B1_BIODIVERSIDADE C_CIÊNCIAS_AGRÁRIAS_I</secondarymark>
		<transferableflag>1</transferableflag>
		<contenttype>External Contribution</contenttype>
		<versiontype>publisher</versiontype>
		<keywords>Change detection, CycleGAN, Deep learning, Deforestation detection, Domain adaptation, Remote sensing.</keywords>
		<abstract>Changes in environmental conditions, geographical variability and different sensor properties typically make it almost impossible to employ previously trained classifiers for new data without a significant drop in classification accuracy. Domain adaptation (DA) techniques been proven useful to alleviate that problem. In particular, appearance adaptation techniques may be used to adapt images from a specific dataset in such a way that the generated images have a style that is similar to the images from another dataset. Such techniques are, however, prone to creating artifacts that hinder proper classification of the adapted images. In this work we propose an unsupervised DA approach for change detection tasks, which is based on a particular appearance adaptation method: the Cycle-Consistent Generative Adversarial Network (CycleGAN). Specifically, we extend that method by introducing additional constraints in the training phase of the model components, which make it preserve the semantic structure and class transitions in the adapted images. We evaluate the proposed approach on a deforestation detection application, considering different sites in the Amazon rain-forest and in the Brazilian Cerrado (savanna) using Landsat-8 images. In the experiments, each site corresponds to a domain, and the accuracy of a classifier trained with images and references from one (source) domain is measured in the classification of another (target) domain. The results show that the proposed approach is successful in producing artifact-free adapted images, which can be satisfactory classified by the pre-trained source classifiers. On average, the accuracies achieved in the classification of the adapted images outperformed the baselines (when no adaptation was made) by 7.1% in terms of mean average precision, and 9.1% in terms of F1-Score. To the best of our knowledge, the proposed method is the first unsupervised domain adaptation approach devised for change detection.</abstract>
		<area>SRE</area>
		<language>en</language>
		<targetfile>vega_unsupervised.pdf</targetfile>
		<usergroup>simone</usergroup>
		<readergroup>administrator</readergroup>
		<readergroup>simone</readergroup>
		<visibility>shown</visibility>
		<archivingpolicy>denypublisher denyfinaldraft24</archivingpolicy>
		<readpermission>deny from all and allow from 150.163</readpermission>
		<documentstage>not transferred</documentstage>
		<nexthigherunit>8JMKD3MGPCW/46L2FGP</nexthigherunit>
		<citingitemlist>sid.inpe.br/bibdigital/2022/04.04.04.47 1</citingitemlist>
		<dissemination>WEBSCI; PORTALCAPES; COMPENDEX; SCOPUS.</dissemination>
		<hostcollection>urlib.net/www/2021/06.04.03.40</hostcollection>
		<username>simone</username>
		<agreement>agreement.html .htaccess .htaccess2</agreement>
		<lasthostcollection>urlib.net/www/2021/06.04.03.40</lasthostcollection>
		<url>http://mtc-m21d.sid.inpe.br/rep-/sid.inpe.br/mtc-m21d/2021/10.04.15.37</url>
	</metadata>
</metadatalist>