Show simple item record

Multi-task deep learning for image segmentation using recursive approximation tasks

hal.structure.identifierDepartment of Applied Mathematics and Theoretical Physics [DAMTP]
dc.contributor.authorKE, Rihuan
hal.structure.identifierLaboratoire Bordelais de Recherche en Informatique [LaBRI]
dc.contributor.authorBUGEAU, Aurélie
hal.structure.identifierInstitut de Mathématiques de Bordeaux [IMB]
dc.contributor.authorPAPADAKIS, Nicolas
dc.contributor.authorKIRKLAND, Mark
dc.contributor.authorSCHUETZ, Peter
hal.structure.identifierDepartment of Applied Mathematics and Theoretical Physics [DAMTP]
dc.contributor.authorSCHÖNLIEB, Carola-Bibiane
dc.date.accessioned2024-04-04T02:54:29Z
dc.date.available2024-04-04T02:54:29Z
dc.date.issued2021
dc.identifier.issn1057-7149
dc.identifier.urihttps://oskar-bordeaux.fr/handle/20.500.12278/192276
dc.description.abstractEnFully supervised deep neural networks for segmentation usually require a massive amount of pixel-level labels which are manually expensive to create. In this work, we develop a multi-task learning method to relax this constraint. We regard the segmentation problem as a sequence of approximation subproblems that are recursively defined and in increasing levels of approximation accuracy. The subproblems are handled by a framework that consists of 1) a segmentation task that learns from pixel-level ground truth segmentation masks of a small fraction of the images, 2) a recursive approximation task that conducts partial object regions learning and data-driven mask evolution starting from partial masks of each object instance, and 3) other problem oriented auxiliary tasks that are trained with sparse annotations and promote the learning of dedicated features. Most training images are only labeled by (rough) partial masks, which do not contain exact object boundaries, rather than by their full segmentation masks. During the training phase, the approximation task learns the statistics of these partial masks, and the partial regions are recursively increased towards object boundaries aided by the learned information from the segmentation task in a fully data-driven fashion. The network is trained on an extremely small amount of precisely segmented images and a large set of coarse labels. Annotations can thus be obtained in a cheap way. We demonstrate the efficiency of our approach in three applications with microscopy images and ultrasound images.
dc.language.isoen
dc.publisherInstitute of Electrical and Electronics Engineers
dc.title.enMulti-task deep learning for image segmentation using recursive approximation tasks
dc.typeArticle de revue
dc.identifier.doi10.1109/TIP.2021.3062726
dc.subject.halInformatique [cs]/Traitement du signal et de l'image
dc.identifier.arxiv2005.13053
dc.description.sponsorshipEuropeNonlocal Methods for Arbitrary Data Sources
bordeaux.journalIEEE Transactions on Image Processing
bordeaux.page3555-3567
bordeaux.volume30
bordeaux.hal.laboratoriesInstitut de Mathématiques de Bordeaux (IMB) - UMR 5251*
bordeaux.institutionUniversité de Bordeaux
bordeaux.institutionBordeaux INP
bordeaux.institutionCNRS
bordeaux.peerReviewedoui
hal.identifierhal-02638394
hal.version1
hal.popularnon
hal.audienceInternationale
hal.origin.linkhttps://hal.archives-ouvertes.fr//hal-02638394v1
bordeaux.COinSctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.jtitle=IEEE%20Transactions%20on%20Image%20Processing&rft.date=2021&rft.volume=30&rft.spage=3555-3567&rft.epage=3555-3567&rft.eissn=1057-7149&rft.issn=1057-7149&rft.au=KE,%20Rihuan&BUGEAU,%20Aur%C3%A9lie&PAPADAKIS,%20Nicolas&KIRKLAND,%20Mark&SCHUETZ,%20Peter&rft.genre=article


Files in this item

FilesSizeFormatView

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record