Efficient crack segmentation with multi-decoder networks and enhanced feature fusion

Okran, Ammar M; Rashwan, Hatem A; Saleh, Adel; Puig, Domenec

Belongs to PC:SerieArticles collection

TITLE:

Efficient crack segmentation with multi-decoder networks and enhanced feature fusion - imarina:9452174

Handle:

https://hdl.handle.net/20.500.11797/imarina9452174

URV's Author/s:	Abdellatif Fatahallah Ibrahim Mahmoud, Hatem / Puig Valls, Domènec Savi
Author, as appears in the article.:	Okran, Ammar M; Rashwan, Hatem A; Saleh, Adel; Puig, Domenec
Author's mail:	domenec.puig@urv.cat hatem.abdellatif@urv.cat
Author identifier:	0000-0002-0562-4205 0000-0001-5421-1637
Journal publication year:	2025
Publication Type:	Journal Publications
APA:	Okran, Ammar M; Rashwan, Hatem A; Saleh, Adel; Puig, Domenec (2025). Efficient crack segmentation with multi-decoder networks and enhanced feature fusion. Engineering Applications Of Artificial Intelligence, 152(), 110697-. DOI: 10.1016/j.engappai.2025.110697
Paper original source:	Engineering Applications Of Artificial Intelligence. 152 110697-
Abstract:	In infrastructure management, intelligent crack detection is vital, particularly for maintaining crucial elements such as road networks in urban areas. Detecting pavement defects promptly and accurately is essential for timely repairs and hazard prevention. However, this task is challenging due to factors, such as complex backgrounds, micro defects, diverse defect shapes and sizes, and class imbalance issues. Innovative approaches and advanced technologies are needed to address these challenges and effectively manage infrastructure complexities. In this study, we propose a novel framework for crack segmentation, called CrackMaster. CrackMaster utilizes advanced neural network architectures, leveraging the next generation of convolutional networks (ConvNeXt) as an encoder and dual decoders customized for distinct tasks. The first decoder adopts a self-supervised learning paradigm to reconstruct images, thereby enhancing feature extraction capabilities. Meanwhile, the second decoder combines deep labelling network for semantic image segmentation (Deeplabv3+) with a light deep neural network (LinkNet) to facilitate precise segmentation. Notably, we introduce an Enhanced Feature Fusion (EFF) block to improve features quality, enhancing information flow and context preservation, thus boosting segmentation performance. Experimental results conducted on three diverse datasets, including our in-house Road Crack Dataset (RCD), DeepCrack537, and Yang Crack Dataset (YCD) datasets, demonstrate the effectiveness of our framework achieving outstanding Intersection over Union scores (IoU) of 86.0%, 87.8%, and 76.9%, respectively, showing superior accuracy and robustness in crack segmentation tasks. These findings underscore the potential applicability of our framework in real-world infrastructure management scenarios. The code is publicly available at: https://github.com/AmmarOkran/ CrackMaster.
Article's DOI:	10.1016/j.engappai.2025.110697
Link to the original source:	https://www.sciencedirect.com/science/article/pii/S0952197625006979?via%3Dihub
Paper version:	info:eu-repo/semantics/publishedVersion
licence for use:	https://creativecommons.org/licenses/by/3.0/es/
Department:	Enginyeria Informàtica i Matemàtiques
Licence document URL:	https://repositori.urv.cat/ca/proteccio-de-dades/
Thematic Areas:	Administração pública e de empresas, ciências contábeis e turismo Artificial intelligence Automation & control systems Biotecnología Ciência da computação Ciência de alimentos Ciências agrárias i Computer science, artificial intelligence Control and systems engineering Electrical and electronic engineering Engenharias i Engenharias ii Engenharias iii Engenharias iv Engineering Engineering, electrical & electronic Engineering, multidisciplinary Interdisciplinar Linguística e literatura Matemática / probabilidade e estatística Materiais Medicina i Robotics & automatic control
Keywords:	Algorith Artificial intelligence Convolutional neural networks Deep learning Pavement crack segmentatio Pavement crack segmentation Road damage detection Self supervised learning Semantic segmentation
Entity:	Universitat Rovira i Virgili
Record's date:	2025-04-30

Description:	In infrastructure management, intelligent crack detection is vital, particularly for maintaining crucial elements such as road networks in urban areas. Detecting pavement defects promptly and accurately is essential for timely repairs and hazard prevention. However, this task is challenging due to factors, such as complex backgrounds, micro defects, diverse defect shapes and sizes, and class imbalance issues. Innovative approaches and advanced technologies are needed to address these challenges and effectively manage infrastructure complexities. In this study, we propose a novel framework for crack segmentation, called CrackMaster. CrackMaster utilizes advanced neural network architectures, leveraging the next generation of convolutional networks (ConvNeXt) as an encoder and dual decoders customized for distinct tasks. The first decoder adopts a self-supervised learning paradigm to reconstruct images, thereby enhancing feature extraction capabilities. Meanwhile, the second decoder combines deep labelling network for semantic image segmentation (Deeplabv3+) with a light deep neural network (LinkNet) to facilitate precise segmentation. Notably, we introduce an Enhanced Feature Fusion (EFF) block to improve features quality, enhancing information flow and context preservation, thus boosting segmentation performance. Experimental results conducted on three diverse datasets, including our in-house Road Crack Dataset (RCD), DeepCrack537, and Yang Crack Dataset (YCD) datasets, demonstrate the effectiveness of our framework achieving outstanding Intersection over Union scores (IoU) of 86.0%, 87.8%, and 76.9%, respectively, showing superior accuracy and robustness in crack segmentation tasks. These findings underscore the potential applicability of our framework in real-world in
Title:	Efficient crack segmentation with multi-decoder networks and enhanced feature fusion
Type:	Journal Publications info:eu-repo/semantics/publishedVersion
Contributor:	Enginyeria Informàtica i Matemàtiques Universitat Rovira i Virgili
Subject:	Artificial Intelligence,Automation & Control Systems,Computer Science, Artificial Intelligence,Control and Systems Engineering,Electrical and Electronic Engineering,Engineering,Engineering, Electrical & Electronic,Engineering, Multidisciplinary,Robotics & Automatic Control Algorith Artificial intelligence Convolutional neural networks Deep learning Pavement crack segmentatio Pavement crack segmentation Road damage detection Self supervised learning Semantic segmentation Administração pública e de empresas, ciências contábeis e turismo Artificial intelligence Automation & control systems Biotecnología Ciência da computação Ciência de alimentos Ciências agrárias i Computer science, artificial intelligence Control and systems engineering Electrical and electronic engineering Engenharias i Engenharias ii Engenharias iii Engenharias iv Engineering Engineering, electrical & electronic Engineering, multidisciplinary Interdisciplinar Linguística e literatura Matemática / probabilidade e estatística Materiais Medicina i Robotics & automatic control
Date:	2025
Language:	en
Creator:	Okran, Ammar M Rashwan, Hatem A Saleh, Adel Puig, Domenec
Rights:	info:eu-repo/semantics/openAccess

Search your record at:

Available files
File	Description	Format
DocumentPrincipal	DocumentPrincipal	application/pdf

All objects of this collection

Information