Show simple item record

SCOTCH and SODA: A Transformer Video Shadow Detection Framework

hal.structure.identifierDepartment of Applied Mathematics and Theoretical Physics [Cambridge] [DAMTP]
dc.contributor.authorLIU, Lihao
hal.structure.identifierInstitut de Mathématiques de Bordeaux [IMB]
dc.contributor.authorPROST, Jean
hal.structure.identifierElectronic and Computer Engineering Department [Hong Kong] [ECE]
dc.contributor.authorZHU, Lei
hal.structure.identifierInstitut de Mathématiques de Bordeaux [IMB]
dc.contributor.authorPAPADAKIS, Nicolas
hal.structure.identifierDepartment of Applied Mathematics and Theoretical Physics [Cambridge] [DAMTP]
dc.contributor.authorLIÒ, Pietro
hal.structure.identifierDepartment of Applied Mathematics and Theoretical Physics [Cambridge] [DAMTP]
dc.contributor.authorSCHÖNLIEB, Carola-Bibiane
hal.structure.identifierDepartment of Applied Mathematics and Theoretical Physics [Cambridge] [DAMTP]
dc.contributor.authorAVILES-RIVERO, Angelica I
dc.date.accessioned2024-04-04T02:35:55Z
dc.date.available2024-04-04T02:35:55Z
dc.date.issued2022-11-13
dc.date.conference2023-06-18
dc.identifier.urihttps://oskar-bordeaux.fr/handle/20.500.12278/190684
dc.description.abstractEnShadows in videos are difficult to detect because of the large shadow deformation between frames. In this work, we argue that accounting for the shadow deformation is essential when designing a video shadow detection method. To this end, we introduce the shadow deformation attention trajectory (SODA), a new type of video self-attention module, specially designed to handle the large shadow deformations in videos. Moreover, we present a shadow contrastive learning mechanism (SCOTCH) which aims at guiding the network to learn a high-level representation of shadows, unified across different videos. We demonstrate empirically the effectiveness of our two contributions in an ablation study. Furthermore, we show that SCOTCH and SODA significantly outperforms existing techniques for video shadow detection. Code will be available upon the acceptance of this work.
dc.language.isoen
dc.title.enSCOTCH and SODA: A Transformer Video Shadow Detection Framework
dc.typeCommunication dans un congrès
dc.subject.halInformatique [cs]/Vision par ordinateur et reconnaissance de formes [cs.CV]
dc.identifier.arxiv2211.06885
dc.description.sponsorshipEuropeNonlocal Methods for Arbitrary Data Sources
bordeaux.hal.laboratoriesInstitut de Mathématiques de Bordeaux (IMB) - UMR 5251*
bordeaux.institutionUniversité de Bordeaux
bordeaux.institutionBordeaux INP
bordeaux.institutionCNRS
bordeaux.conference.titleIEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR'23)
bordeaux.countryCA
bordeaux.conference.cityVancouver
bordeaux.peerReviewedoui
hal.identifierhal-03944252
hal.version1
hal.proceedingsoui
hal.conference.end2023-06-22
hal.popularnon
hal.audienceInternationale
hal.origin.linkhttps://hal.archives-ouvertes.fr//hal-03944252v1
bordeaux.COinSctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.date=2022-11-13&rft.au=LIU,%20Lihao&PROST,%20Jean&ZHU,%20Lei&PAPADAKIS,%20Nicolas&LI%C3%92,%20Pietro&rft.genre=unknown


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