Automatic learning of image representations combining content and metadata

e-Archivo Repository

Show simple item record

dc.contributor.author Martínez Cortés, Tomas
dc.contributor.author González Díaz, Iván
dc.contributor.author Díaz de María, Fernando
dc.date.accessioned 2020-09-24T10:45:40Z
dc.date.available 2020-09-24T10:45:40Z
dc.date.issued 2018-10-07
dc.identifier.bibliographicCitation 2018 IEEE International Conference on Image Processing: proceedings: October 7-10, 2018, Athens, Greece. USA: IEEE, 2018. Pp. 1972-1976
dc.identifier.isbn 978-1-4799-7061-2
dc.identifier.uri http://hdl.handle.net/10016/30849
dc.description Proceeding of: 25th IEEE International Conference on Image Processing (ICIP 2018 ), 7-10 October, 2018, Athens, Greece
dc.description.abstract Content-based image representation is a very challenging task if we restrict to their visual content. However, associated metadata (such as tags or geolocation) become a valuable source of complementary information that may help to enhance the current system performance. In this paper, we propose an automatic training framework that uses both image visual contents and metadata to fine tune deep Convolutional Neural Networks (CNNs), providing better image descriptors adapted to certain locations, such as cities or regions. Specifically, we propose to estimate some weak labels by combining visual- and location-related information and incorporate them to a novel loss-function over pairs of images. Our experiments on a landmark discovery task show that this novel training procedure enhances the performance up to a 55% over well-established CNN-based models and is free from overfitting
dc.description.sponsorship This work has been partially supported by the National Grants TEC2014-53390-P and TEC2017-84395-P of the Spanish Ministry of Economy and Competitiveness.
dc.format.extent 4
dc.language.iso eng
dc.publisher IEEE
dc.rights ©2018 Crown
dc.rights ©2018 by The Institute of Electrical and Electronics Engineers, Inc. All rights reserved.
dc.subject.other CNN
dc.subject.other Metadata
dc.subject.other Loss function
dc.subject.other Weak labels
dc.title Automatic learning of image representations combining content and metadata
dc.type conferenceObject
dc.description.status Publicado
dc.subject.eciencia Telecomunicaciones
dc.identifier.doi https://doi.org/10.1109/ICIP.2018.8451566
dc.rights.accessRights openAccess
dc.relation.projectID Gobierno de España. TEC2014-53390-P
dc.relation.projectID Gobierno de España. TEC2017-84395-P
dc.type.version acceptedVersion
dc.relation.eventdate 7-10 October, 2018
dc.relation.eventnumber 25
dc.relation.eventplace Athens, Greece
dc.relation.eventtitle IEEE International Conference on Image Processing (ICIP)
dc.relation.eventtype proceeding
dc.identifier.publicationfirstpage 1972
dc.identifier.publicationlastpage 1976
dc.identifier.publicationtitle 2018 IEEE International Conference on Image Processing: proceedings: October 7-10, 2018, Athens, Greece.
dc.identifier.uxxi CC/0000029122
dc.contributor.funder Ministerio de Economía y Competitividad (España)
 Find Full text

Files in this item

*Click on file's image for preview. (Embargoed files's preview is not supported)


This item appears in the following Collection(s)

Show simple item record