Shadow detection using a cross-attentional dual-decoder network with self-supervised image reconstruction features

Fernández Beltrán, Rubén; Guzmán Ponce, Angélica; Fernandez, Rafael; Kang, Jian; García Mateos, Ginés

Publication:
Shadow detection using a cross-attentional dual-decoder network with self-supervised image reconstruction features

dc.contributor.author	Fernández Beltrán, Rubén
dc.contributor.author	Guzmán Ponce, Angélica
dc.contributor.author	Fernandez, Rafael
dc.contributor.author	Kang, Jian
dc.contributor.author	García Mateos, Ginés
dc.contributor.department	Informática y Sistemas
dc.contributor.other	Facultad de Informática
dc.date.accessioned	2026-01-19T08:59:39Z
dc.date.available	2026-01-19T08:59:39Z
dc.date.copyright	© 2024 The Author(s)
dc.date.copyright	© 2024 The Author(s)
dc.date.issued	2024-02-02
dc.description.abstract	Shadow detection is a challenging problem in computer vision due to the high variability in lighting conditions, object shapes, and scene layouts. Despite the positive results achieved by some existing technologies, the problem becomes particularly challenging with complex and heterogeneous images where shadow-casting objects coexist and shadows can have different depths, scales, and morphologies. As a result, more advanced and accurate solutions are still needed to deal with this type of complexities. To address these challenges, this paper proposes a novel deep learning model, called the Cross-Attentional Dual Decoder Network (CADDN), to improve shadow detection by using fine-grained image reconstruction features. Unlike other existing methods, the CADDN uses an innovative encoder-decoder architecture with two decoder segments that work together to reconstruct the input images and their corresponding shadow masks. In this way, the features used to reconstruct the original input image can be used to support the shadow detection process itself. The proposed model also incorporates a cross-attention mechanism to weight the most relevant features for detecting shadows and skip connections with noise to improve the quality of the transferred features. The experimental results, including several benchmark image datasets and state-of-the-art detection methods, demonstrate the suitability of the presented approach for detecting shadows in computer vision applications.
dc.format	application/pdf
dc.format.extent	16
dc.identifier.citation	Image and Vision Computing 143 (2024) 104922
dc.identifier.doi	https://doi.org/10.1016/j.imavis.2024.104922
dc.identifier.eissn	1872‑8138
dc.identifier.issn	0262‑8856
dc.identifier.uri	http://hdl.handle.net/10201/187990
dc.language	eng
dc.publisher	Elsevier
dc.relation	This work was supported by the Postdoctoral Margarita Salas Fellowship MGS/2021/23(UP2021-021) from the European Union-NextGenerationEU funds, the Science and Technology Agency of the Region of Murcia (Fundación Séneca, Action Plan 2022) by grant 22130/PI/22, and the National Natural Science Foundation of China under Grant 62101371.
dc.relation.publisherversion	https://www.sciencedirect.com/science/article/pii/S0262885624000258
dc.rights	Attribution-NonCommercial-NoDerivatives 4.0 International	*
dc.rights.accessRights	info:eu-repo/semantics/openAccess
dc.rights.uri	http://creativecommons.org/licenses/by-nc-nd/4.0/	*
dc.subject	Shadow detection
dc.subject	Semantic segmentation
dc.subject	Convolutional neural networks
dc.subject	Cross attention
dc.subject	Dual decoder
dc.subject.ods	Objetivo 7: Energía
dc.subject.ods	Objetivo 9: Infraestructura
dc.title	Shadow detection using a cross-attentional dual-decoder network with self-supervised image reconstruction features
dc.type	info:eu-repo/semantics/article
dc.type.version	info:eu-repo/semantics/publishedVersion
dspace.entity.type	Publication	es
relation.isAuthorOfPublication	4b1f10dc-5b0a-4f2b-b544-9b5a81e3a07e
relation.isAuthorOfPublication	bb68ac7e-f852-4bc6-a7b4-e9a37a5617c2
relation.isAuthorOfPublication.latestForDiscovery	4b1f10dc-5b0a-4f2b-b544-9b5a81e3a07e