A Multi-Scale attention network for building extraction from high-resolution remote sensing images

Abstract The information in remote sensing images often leads to incomplete building contours and suboptimal adaptability to complex building scenes. To address these issues, we propose a novel multi-scale network with dual attention mechanisms to extract clear building boundaries. The Squeeze-and-E...

Full description

Saved in:
Bibliographic Details
Main Authors: Jing Chang, Xiaohui He, Dingjun Song, Panle Li, Mengjia Qiao, Xijie Cheng
Format: Article
Language:English
Published: Nature Portfolio 2025-07-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-09086-9
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849388011543855104
author Jing Chang
Xiaohui He
Dingjun Song
Panle Li
Mengjia Qiao
Xijie Cheng
author_facet Jing Chang
Xiaohui He
Dingjun Song
Panle Li
Mengjia Qiao
Xijie Cheng
author_sort Jing Chang
collection DOAJ
description Abstract The information in remote sensing images often leads to incomplete building contours and suboptimal adaptability to complex building scenes. To address these issues, we propose a novel multi-scale network with dual attention mechanisms to extract clear building boundaries. The Squeeze-and-Excitation (SE) module is employed to bolster feature extraction, and the Atrous Spatial Pyramid Pooling (ASPP) module is integrated to capture multi-scale feature information. Then, in the decoding phase, channel grouping shuffle and dual attention mechanisms are synergistically integrated to exploit the interrelations and global dependencies of building features. Finally, a hybrid loss function is devised to address the class imbalance and thereby ensure more stable network training. Experimental evaluations on two high-resolution remote sensing datasets, Zimbabwe and Massachusetts, demonstrate that the proposed method markedly surpasses the performance of semantic segmentation networks such as PSPnet, U-net, and DAnet in terms of accuracy, recall, F1 score, and Mean Intersection over Union (MIoU), achieving an F1 score of up to 83.23% and an MIoU of 73.56%. This multi-scale attention network holds substantial promise for practical applications in building extraction.
format Article
id doaj-art-a90286c434db435e87fdb4af09995564
institution Kabale University
issn 2045-2322
language English
publishDate 2025-07-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-a90286c434db435e87fdb4af099955642025-08-20T03:42:25ZengNature PortfolioScientific Reports2045-23222025-07-0115111510.1038/s41598-025-09086-9A Multi-Scale attention network for building extraction from high-resolution remote sensing imagesJing Chang0Xiaohui He1Dingjun Song2Panle Li3Mengjia Qiao4Xijie Cheng5School of Computer and Artificial Intelligence, Zhengzhou UniversitySchool of Geoscience and Technology, Zhengzhou UniversitySchool of Computer and Artificial Intelligence, Zhengzhou UniversitySchool of Geoscience and Technology, Zhengzhou UniversitySchool of Geoscience and Technology, Zhengzhou UniversitySchool of Geoscience and Technology, Zhengzhou UniversityAbstract The information in remote sensing images often leads to incomplete building contours and suboptimal adaptability to complex building scenes. To address these issues, we propose a novel multi-scale network with dual attention mechanisms to extract clear building boundaries. The Squeeze-and-Excitation (SE) module is employed to bolster feature extraction, and the Atrous Spatial Pyramid Pooling (ASPP) module is integrated to capture multi-scale feature information. Then, in the decoding phase, channel grouping shuffle and dual attention mechanisms are synergistically integrated to exploit the interrelations and global dependencies of building features. Finally, a hybrid loss function is devised to address the class imbalance and thereby ensure more stable network training. Experimental evaluations on two high-resolution remote sensing datasets, Zimbabwe and Massachusetts, demonstrate that the proposed method markedly surpasses the performance of semantic segmentation networks such as PSPnet, U-net, and DAnet in terms of accuracy, recall, F1 score, and Mean Intersection over Union (MIoU), achieving an F1 score of up to 83.23% and an MIoU of 73.56%. This multi-scale attention network holds substantial promise for practical applications in building extraction.https://doi.org/10.1038/s41598-025-09086-9Deep learningResidual moduleDual attention mechanismMulti-Scale feature integration
spellingShingle Jing Chang
Xiaohui He
Dingjun Song
Panle Li
Mengjia Qiao
Xijie Cheng
A Multi-Scale attention network for building extraction from high-resolution remote sensing images
Scientific Reports
Deep learning
Residual module
Dual attention mechanism
Multi-Scale feature integration
title A Multi-Scale attention network for building extraction from high-resolution remote sensing images
title_full A Multi-Scale attention network for building extraction from high-resolution remote sensing images
title_fullStr A Multi-Scale attention network for building extraction from high-resolution remote sensing images
title_full_unstemmed A Multi-Scale attention network for building extraction from high-resolution remote sensing images
title_short A Multi-Scale attention network for building extraction from high-resolution remote sensing images
title_sort multi scale attention network for building extraction from high resolution remote sensing images
topic Deep learning
Residual module
Dual attention mechanism
Multi-Scale feature integration
url https://doi.org/10.1038/s41598-025-09086-9
work_keys_str_mv AT jingchang amultiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT xiaohuihe amultiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT dingjunsong amultiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT panleli amultiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT mengjiaqiao amultiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT xijiecheng amultiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT jingchang multiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT xiaohuihe multiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT dingjunsong multiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT panleli multiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT mengjiaqiao multiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages
AT xijiecheng multiscaleattentionnetworkforbuildingextractionfromhighresolutionremotesensingimages