基于深度学习的语义分割平屋顶污点映射

IF 6.5 2区 工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY
Lara Monalisa Alves dos Santos , Leonardo Rabero Lescano , Gabriel Toshio Hirokawa Higa , Vanda Alice Garcia Zanoni , Lenildo Santos da Silva , Cesar Ivan Alvarez , Hemerson Pistori
{"title":"基于深度学习的语义分割平屋顶污点映射","authors":"Lara Monalisa Alves dos Santos ,&nbsp;Leonardo Rabero Lescano ,&nbsp;Gabriel Toshio Hirokawa Higa ,&nbsp;Vanda Alice Garcia Zanoni ,&nbsp;Lenildo Santos da Silva ,&nbsp;Cesar Ivan Alvarez ,&nbsp;Hemerson Pistori","doi":"10.1016/j.cscm.2024.e04106","DOIUrl":null,"url":null,"abstract":"<div><div>Moisture stains indicate ongoing degradation processes and may reveal areas of the roof slab where water infiltration occurs, compromising the performance and durability of the building system. During inspections of roofing systems, an inspector's field of vision differs from that of drones during overflights. As a result, traditional inspections might not always detect the presence and severity of stains, making maintenance on flat roofs a complex task. In this context, this experimental study aims to analyze deep learning-based semantic segmentation with images obtained from drones to map and monitor damp patches during automated building inspections of flat roof systems. The research tested two convolutional neural networks for semantic segmentation: the Fully Convolutional Network (FCN) with a ResNet50 backbone and DeepLabV3 with a ResNet101 backbone, as well as a transformer-based deep artificial neural network called SegFormer with a MiT-B1 backbone. We evaluated three optimizers for each model—Adam, Adagrad, and SGD—along with learning rates of 1e-2, 1e-3, and 1e-4. The models were compared using four performance metrics. The FCN, optimized with Adagrad at a learning rate of 1e-2, showed the best results. The average metrics obtained in this case were as follows: precision: 79.69 %, recall: 67.81 %, F-score: 73.09 %, and Intersection over Union (IoU): 57.70 %.</div></div>","PeriodicalId":9641,"journal":{"name":"Case Studies in Construction Materials","volume":"22 ","pages":"Article e04106"},"PeriodicalIF":6.5000,"publicationDate":"2024-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Mapping stains on flat roofs using semantic segmentation based on deep learning\",\"authors\":\"Lara Monalisa Alves dos Santos ,&nbsp;Leonardo Rabero Lescano ,&nbsp;Gabriel Toshio Hirokawa Higa ,&nbsp;Vanda Alice Garcia Zanoni ,&nbsp;Lenildo Santos da Silva ,&nbsp;Cesar Ivan Alvarez ,&nbsp;Hemerson Pistori\",\"doi\":\"10.1016/j.cscm.2024.e04106\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Moisture stains indicate ongoing degradation processes and may reveal areas of the roof slab where water infiltration occurs, compromising the performance and durability of the building system. During inspections of roofing systems, an inspector's field of vision differs from that of drones during overflights. As a result, traditional inspections might not always detect the presence and severity of stains, making maintenance on flat roofs a complex task. In this context, this experimental study aims to analyze deep learning-based semantic segmentation with images obtained from drones to map and monitor damp patches during automated building inspections of flat roof systems. The research tested two convolutional neural networks for semantic segmentation: the Fully Convolutional Network (FCN) with a ResNet50 backbone and DeepLabV3 with a ResNet101 backbone, as well as a transformer-based deep artificial neural network called SegFormer with a MiT-B1 backbone. We evaluated three optimizers for each model—Adam, Adagrad, and SGD—along with learning rates of 1e-2, 1e-3, and 1e-4. The models were compared using four performance metrics. The FCN, optimized with Adagrad at a learning rate of 1e-2, showed the best results. The average metrics obtained in this case were as follows: precision: 79.69 %, recall: 67.81 %, F-score: 73.09 %, and Intersection over Union (IoU): 57.70 %.</div></div>\",\"PeriodicalId\":9641,\"journal\":{\"name\":\"Case Studies in Construction Materials\",\"volume\":\"22 \",\"pages\":\"Article e04106\"},\"PeriodicalIF\":6.5000,\"publicationDate\":\"2024-12-18\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Case Studies in Construction Materials\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S2214509524012580\",\"RegionNum\":2,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"CONSTRUCTION & BUILDING TECHNOLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Case Studies in Construction Materials","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2214509524012580","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CONSTRUCTION & BUILDING TECHNOLOGY","Score":null,"Total":0}
引用次数: 0

摘要

湿渍表明正在进行的降解过程,可能会暴露出屋顶板上发生渗水的区域,从而影响建筑系统的性能和耐久性。在检查屋顶系统时,检查员的视野与飞越时的无人机不同。因此,传统的检查可能并不总能检测到污渍的存在和严重程度,这使得平屋顶的维护成为一项复杂的任务。在此背景下,本实验研究旨在分析基于深度学习的语义分割,利用无人机获得的图像来绘制和监测平屋顶系统自动建筑检查过程中的潮湿斑块。该研究测试了两种用于语义分割的卷积神经网络:具有ResNet50骨干网的全卷积网络(FCN)和具有ResNet101骨干网的DeepLabV3,以及具有MiT-B1骨干网的基于变压器的深度人工神经网络SegFormer。我们评估了每个模型的三个优化器——adam、Adagrad和sgd——以及1e-2、1e-3和1e-4的学习率。使用四个性能指标对模型进行比较。使用Adagrad优化的FCN以1e-2的学习率表现出最好的效果。在这种情况下获得的平均指标如下:精度:79.69 %,召回率:67.81 %,f分数:73.09 %,路口/联盟(IoU): 57.70 %。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Mapping stains on flat roofs using semantic segmentation based on deep learning
Moisture stains indicate ongoing degradation processes and may reveal areas of the roof slab where water infiltration occurs, compromising the performance and durability of the building system. During inspections of roofing systems, an inspector's field of vision differs from that of drones during overflights. As a result, traditional inspections might not always detect the presence and severity of stains, making maintenance on flat roofs a complex task. In this context, this experimental study aims to analyze deep learning-based semantic segmentation with images obtained from drones to map and monitor damp patches during automated building inspections of flat roof systems. The research tested two convolutional neural networks for semantic segmentation: the Fully Convolutional Network (FCN) with a ResNet50 backbone and DeepLabV3 with a ResNet101 backbone, as well as a transformer-based deep artificial neural network called SegFormer with a MiT-B1 backbone. We evaluated three optimizers for each model—Adam, Adagrad, and SGD—along with learning rates of 1e-2, 1e-3, and 1e-4. The models were compared using four performance metrics. The FCN, optimized with Adagrad at a learning rate of 1e-2, showed the best results. The average metrics obtained in this case were as follows: precision: 79.69 %, recall: 67.81 %, F-score: 73.09 %, and Intersection over Union (IoU): 57.70 %.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
CiteScore
7.60
自引率
19.40%
发文量
842
审稿时长
63 days
期刊介绍: Case Studies in Construction Materials provides a forum for the rapid publication of short, structured Case Studies on construction materials. In addition, the journal also publishes related Short Communications, Full length research article and Comprehensive review papers (by invitation). The journal will provide an essential compendium of case studies for practicing engineers, designers, researchers and other practitioners who are interested in all aspects construction materials. The journal will publish new and novel case studies, but will also provide a forum for the publication of high quality descriptions of classic construction material problems and solutions.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信