File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Article: A Decoupled Multi-Task Network for Shadow Removal

TitleA Decoupled Multi-Task Network for Shadow Removal
Authors
Keywordsdecomposition
Estimation
feature decoupling
Feature extraction
illumination compensation
Image decomposition
Image restoration
Lighting
multi-task
Multitasking
Shadow removal
Task analysis
Issue Date3-Mar-2023
PublisherInstitute of Electrical and Electronics Engineers
Citation
IEEE Transactions on Multimedia, 2023 How to Cite?
Abstract

Shadow removal, which aims to restore the illumination in shadow regions, is challenging due to the diversity of shadows in terms of location, intensity, shape, and size. Different from most multi-task methods, which design elaborate multi-branch or multi-stage structures for better shadow removal, we introduce feature decomposition to learn better feature representations. Specifically, we propose a single-stage and decoupled multi-task network (DMTN) to explicitly learn the decomposed features for shadow removal, shadow matte estimation, and shadow image reconstruction. First, we propose several coarse-to-fine semi-convolution (SMC) modules to capture features sufficient for joint learning of these three tasks. Second, we design a theoretically supported feature decoupling layer to explicitly decouple the learned features into shadow image features and shadow matte features via weight reassignment. Last, these features are converted to a target shadow-free image, affiliated shadow matte, and shadow image, supervised by multi-task joint loss functions. With multi-task collaboration, DMTN effectively recovers the illumination in shadow areas while ensuring the fidelity of non-shadow areas. Experimental results show that DMTN competes favorably with state-of-the-art multi-branch/multi-stage shadow removal methods, while maintaining the simplicity of single-stage methods. We have released our code to encourage future exploration in powerful feature representation for shadow removal <uri>https://github.com/nachifur/DMTN</uri>


Persistent Identifierhttp://hdl.handle.net/10722/338113
ISSN
2023 Impact Factor: 8.4
2023 SCImago Journal Rankings: 2.260
ISI Accession Number ID

 

DC FieldValueLanguage
dc.contributor.authorLiu, Jiawei-
dc.contributor.authorWang, Qiang-
dc.contributor.authorFan, Huijie-
dc.contributor.authorLi, Wentao-
dc.contributor.authorQu, Liangqiong-
dc.contributor.authorTang, Yandong-
dc.date.accessioned2024-03-11T10:26:21Z-
dc.date.available2024-03-11T10:26:21Z-
dc.date.issued2023-03-03-
dc.identifier.citationIEEE Transactions on Multimedia, 2023-
dc.identifier.issn1520-9210-
dc.identifier.urihttp://hdl.handle.net/10722/338113-
dc.description.abstract<p>Shadow removal, which aims to restore the illumination in shadow regions, is challenging due to the diversity of shadows in terms of location, intensity, shape, and size. Different from most multi-task methods, which design elaborate multi-branch or multi-stage structures for better shadow removal, we introduce feature decomposition to learn better feature representations. Specifically, we propose a single-stage and decoupled multi-task network (DMTN) to explicitly learn the decomposed features for shadow removal, shadow matte estimation, and shadow image reconstruction. First, we propose several coarse-to-fine semi-convolution (SMC) modules to capture features sufficient for joint learning of these three tasks. Second, we design a theoretically supported feature decoupling layer to explicitly decouple the learned features into shadow image features and shadow matte features via weight reassignment. Last, these features are converted to a target shadow-free image, affiliated shadow matte, and shadow image, supervised by multi-task joint loss functions. With multi-task collaboration, DMTN effectively recovers the illumination in shadow areas while ensuring the fidelity of non-shadow areas. Experimental results show that DMTN competes favorably with state-of-the-art multi-branch/multi-stage shadow removal methods, while maintaining the simplicity of single-stage methods. We have released our code to encourage future exploration in powerful feature representation for shadow removal &lt;uri&gt;https://github.com/nachifur/DMTN&lt;/uri&gt;</p>-
dc.languageeng-
dc.publisherInstitute of Electrical and Electronics Engineers-
dc.relation.ispartofIEEE Transactions on Multimedia-
dc.rightsThis work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.-
dc.subjectdecomposition-
dc.subjectEstimation-
dc.subjectfeature decoupling-
dc.subjectFeature extraction-
dc.subjectillumination compensation-
dc.subjectImage decomposition-
dc.subjectImage restoration-
dc.subjectLighting-
dc.subjectmulti-task-
dc.subjectMultitasking-
dc.subjectShadow removal-
dc.subjectTask analysis-
dc.titleA Decoupled Multi-Task Network for Shadow Removal-
dc.typeArticle-
dc.identifier.doi10.1109/TMM.2023.3252271-
dc.identifier.scopuseid_2-s2.0-85149469495-
dc.identifier.eissn1941-0077-
dc.identifier.isiWOS:001133324200038-
dc.identifier.issnl1520-9210-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats