Automatic sewer defect detection and severity quantification based on pixel-level semantic segmentation

•Automated pixel-level sewer defects detection and segmentation.•DeepLabv3+ with five types of backbones are compared with FCN, U-Net and SegNet.•Severity quantification by means of geometric assessment of sewer defects.•Multiple types of sewer defects under complex conditions.•DeepLabv3+-Resnet50 a...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Tunnelling and underground space technology 2022-05, Vol.123, p.104403, Article 104403
Hauptverfasser: Zhou, Qianqian, Situ, Zuxiang, Teng, Shuai, Liu, Hanlin, Chen, Weifeng, Chen, Gongfa
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page 104403
container_title Tunnelling and underground space technology
container_volume 123
creator Zhou, Qianqian
Situ, Zuxiang
Teng, Shuai
Liu, Hanlin
Chen, Weifeng
Chen, Gongfa
description •Automated pixel-level sewer defects detection and segmentation.•DeepLabv3+ with five types of backbones are compared with FCN, U-Net and SegNet.•Severity quantification by means of geometric assessment of sewer defects.•Multiple types of sewer defects under complex conditions.•DeepLabv3+-Resnet50 achieved highest PA, mIoU and F1-score, and fastest speed. Underground sewer systems are characterized by their large-scale distributions, high coverage densities and complex defect conditions, which put forward higher requirements for system inspection. Conventional manual inspection is labor-intensive, error-prone and cost inefficient. This paper presents a novel DeepLabv3+-based sewer defect detection and severity quantification method to facilitate automated pixel-level segmentation of sewer defects, from which the defect types, locations, geometric properties and severity levels can be assessed. The effects of different backbone networks on DeepLabv3+ detection accuracy and processing speed were investigated. Three state-of-the-art segmentation methods (SegNet, FCN and U-Net) were further compared to confirm the feasibility of the proposed method. Our results showed that the DeepLabv3+ model, especially with the backbone network Resnet-50, was superior in segmenting multiple types of sewer defects under complex conditions. The obtained PA, mIoU, fwIoU and F1-score were 0.9, 0.53, 0.84 and 0.55, respectively. For individual types of defects, the model worked best in identifying the type of residential walls, followed by disjoints and tree roots. In terms of severity quantification, it was shown that 70% of model predictions were consistent with the ground truths, whereas 30% of predictions likely overestimated the severity levels. The proposed method provides decision-making basis for more accurate and effective sewer inspection and maintenance.
doi_str_mv 10.1016/j.tust.2022.104403
format Article
fullrecord <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_journals_2647396586</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S0886779822000438</els_id><sourcerecordid>2647396586</sourcerecordid><originalsourceid>FETCH-LOGICAL-c328t-719172cf9e03b2f79bb753aceb9e428b9b3b2a30d80cd90c67bb05c939fd94773</originalsourceid><addsrcrecordid>eNp9kMtOwzAQRS0EEqXwA6wisU5x7CS2JTZVxUuqxAbWlu1MKkd5tLZT6N_jNKxZ3dHMuTOji9B9hlcZzsrHZhVGH1YEExIbeY7pBVpknPE0p2V-iRaY8zJlTPBrdON9gzEuCBELtFuPYehUsCbx8A0uqaAGE6KEKHboE9VXcXQEZ8MpOYyqD7a2Rp1nWnmokljs7Q-0aRuxNsLdBE0Ldx304YzeoqtatR7u_nSJvl6ePzdv6fbj9X2z3qaGEh5SlomMEVMLwFSTmgmtWUGVAS0gJ1wLHduK4opjUwlsSqY1Loygoq5Ezhhdood5794NhxF8kM0wuj6elKTMGRVlwctIkZkybvDeQS33znbKnWSG5RSobOQUqJwClXOg0fQ0myD-f7TgpDcWegOVdTErWQ32P_svlh6BQA</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2647396586</pqid></control><display><type>article</type><title>Automatic sewer defect detection and severity quantification based on pixel-level semantic segmentation</title><source>Access via ScienceDirect (Elsevier)</source><creator>Zhou, Qianqian ; Situ, Zuxiang ; Teng, Shuai ; Liu, Hanlin ; Chen, Weifeng ; Chen, Gongfa</creator><creatorcontrib>Zhou, Qianqian ; Situ, Zuxiang ; Teng, Shuai ; Liu, Hanlin ; Chen, Weifeng ; Chen, Gongfa</creatorcontrib><description>•Automated pixel-level sewer defects detection and segmentation.•DeepLabv3+ with five types of backbones are compared with FCN, U-Net and SegNet.•Severity quantification by means of geometric assessment of sewer defects.•Multiple types of sewer defects under complex conditions.•DeepLabv3+-Resnet50 achieved highest PA, mIoU and F1-score, and fastest speed. Underground sewer systems are characterized by their large-scale distributions, high coverage densities and complex defect conditions, which put forward higher requirements for system inspection. Conventional manual inspection is labor-intensive, error-prone and cost inefficient. This paper presents a novel DeepLabv3+-based sewer defect detection and severity quantification method to facilitate automated pixel-level segmentation of sewer defects, from which the defect types, locations, geometric properties and severity levels can be assessed. The effects of different backbone networks on DeepLabv3+ detection accuracy and processing speed were investigated. Three state-of-the-art segmentation methods (SegNet, FCN and U-Net) were further compared to confirm the feasibility of the proposed method. Our results showed that the DeepLabv3+ model, especially with the backbone network Resnet-50, was superior in segmenting multiple types of sewer defects under complex conditions. The obtained PA, mIoU, fwIoU and F1-score were 0.9, 0.53, 0.84 and 0.55, respectively. For individual types of defects, the model worked best in identifying the type of residential walls, followed by disjoints and tree roots. In terms of severity quantification, it was shown that 70% of model predictions were consistent with the ground truths, whereas 30% of predictions likely overestimated the severity levels. The proposed method provides decision-making basis for more accurate and effective sewer inspection and maintenance.</description><identifier>ISSN: 0886-7798</identifier><identifier>EISSN: 1878-4364</identifier><identifier>DOI: 10.1016/j.tust.2022.104403</identifier><language>eng</language><publisher>Oxford: Elsevier Ltd</publisher><subject>Computer networks ; Decision making ; DeepLabv3 ; Defects ; Inspection ; Pixel level ; Pixels ; Repair &amp; maintenance ; Semantic segmentation ; Sensors ; Severity quantification ; Sewer defect detection ; Sewer maintenance ; Sewer systems</subject><ispartof>Tunnelling and underground space technology, 2022-05, Vol.123, p.104403, Article 104403</ispartof><rights>2022 Elsevier Ltd</rights><rights>Copyright Elsevier BV May 2022</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c328t-719172cf9e03b2f79bb753aceb9e428b9b3b2a30d80cd90c67bb05c939fd94773</citedby><cites>FETCH-LOGICAL-c328t-719172cf9e03b2f79bb753aceb9e428b9b3b2a30d80cd90c67bb05c939fd94773</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://dx.doi.org/10.1016/j.tust.2022.104403$$EHTML$$P50$$Gelsevier$$H</linktohtml><link.rule.ids>314,780,784,3550,27924,27925,45995</link.rule.ids></links><search><creatorcontrib>Zhou, Qianqian</creatorcontrib><creatorcontrib>Situ, Zuxiang</creatorcontrib><creatorcontrib>Teng, Shuai</creatorcontrib><creatorcontrib>Liu, Hanlin</creatorcontrib><creatorcontrib>Chen, Weifeng</creatorcontrib><creatorcontrib>Chen, Gongfa</creatorcontrib><title>Automatic sewer defect detection and severity quantification based on pixel-level semantic segmentation</title><title>Tunnelling and underground space technology</title><description>•Automated pixel-level sewer defects detection and segmentation.•DeepLabv3+ with five types of backbones are compared with FCN, U-Net and SegNet.•Severity quantification by means of geometric assessment of sewer defects.•Multiple types of sewer defects under complex conditions.•DeepLabv3+-Resnet50 achieved highest PA, mIoU and F1-score, and fastest speed. Underground sewer systems are characterized by their large-scale distributions, high coverage densities and complex defect conditions, which put forward higher requirements for system inspection. Conventional manual inspection is labor-intensive, error-prone and cost inefficient. This paper presents a novel DeepLabv3+-based sewer defect detection and severity quantification method to facilitate automated pixel-level segmentation of sewer defects, from which the defect types, locations, geometric properties and severity levels can be assessed. The effects of different backbone networks on DeepLabv3+ detection accuracy and processing speed were investigated. Three state-of-the-art segmentation methods (SegNet, FCN and U-Net) were further compared to confirm the feasibility of the proposed method. Our results showed that the DeepLabv3+ model, especially with the backbone network Resnet-50, was superior in segmenting multiple types of sewer defects under complex conditions. The obtained PA, mIoU, fwIoU and F1-score were 0.9, 0.53, 0.84 and 0.55, respectively. For individual types of defects, the model worked best in identifying the type of residential walls, followed by disjoints and tree roots. In terms of severity quantification, it was shown that 70% of model predictions were consistent with the ground truths, whereas 30% of predictions likely overestimated the severity levels. The proposed method provides decision-making basis for more accurate and effective sewer inspection and maintenance.</description><subject>Computer networks</subject><subject>Decision making</subject><subject>DeepLabv3</subject><subject>Defects</subject><subject>Inspection</subject><subject>Pixel level</subject><subject>Pixels</subject><subject>Repair &amp; maintenance</subject><subject>Semantic segmentation</subject><subject>Sensors</subject><subject>Severity quantification</subject><subject>Sewer defect detection</subject><subject>Sewer maintenance</subject><subject>Sewer systems</subject><issn>0886-7798</issn><issn>1878-4364</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><recordid>eNp9kMtOwzAQRS0EEqXwA6wisU5x7CS2JTZVxUuqxAbWlu1MKkd5tLZT6N_jNKxZ3dHMuTOji9B9hlcZzsrHZhVGH1YEExIbeY7pBVpknPE0p2V-iRaY8zJlTPBrdON9gzEuCBELtFuPYehUsCbx8A0uqaAGE6KEKHboE9VXcXQEZ8MpOYyqD7a2Rp1nWnmokljs7Q-0aRuxNsLdBE0Ldx304YzeoqtatR7u_nSJvl6ePzdv6fbj9X2z3qaGEh5SlomMEVMLwFSTmgmtWUGVAS0gJ1wLHduK4opjUwlsSqY1Loygoq5Ezhhdood5794NhxF8kM0wuj6elKTMGRVlwctIkZkybvDeQS33znbKnWSG5RSobOQUqJwClXOg0fQ0myD-f7TgpDcWegOVdTErWQ32P_svlh6BQA</recordid><startdate>202205</startdate><enddate>202205</enddate><creator>Zhou, Qianqian</creator><creator>Situ, Zuxiang</creator><creator>Teng, Shuai</creator><creator>Liu, Hanlin</creator><creator>Chen, Weifeng</creator><creator>Chen, Gongfa</creator><general>Elsevier Ltd</general><general>Elsevier BV</general><scope>AAYXX</scope><scope>CITATION</scope><scope>8FD</scope><scope>FR3</scope><scope>KR7</scope></search><sort><creationdate>202205</creationdate><title>Automatic sewer defect detection and severity quantification based on pixel-level semantic segmentation</title><author>Zhou, Qianqian ; Situ, Zuxiang ; Teng, Shuai ; Liu, Hanlin ; Chen, Weifeng ; Chen, Gongfa</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c328t-719172cf9e03b2f79bb753aceb9e428b9b3b2a30d80cd90c67bb05c939fd94773</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Computer networks</topic><topic>Decision making</topic><topic>DeepLabv3</topic><topic>Defects</topic><topic>Inspection</topic><topic>Pixel level</topic><topic>Pixels</topic><topic>Repair &amp; maintenance</topic><topic>Semantic segmentation</topic><topic>Sensors</topic><topic>Severity quantification</topic><topic>Sewer defect detection</topic><topic>Sewer maintenance</topic><topic>Sewer systems</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Zhou, Qianqian</creatorcontrib><creatorcontrib>Situ, Zuxiang</creatorcontrib><creatorcontrib>Teng, Shuai</creatorcontrib><creatorcontrib>Liu, Hanlin</creatorcontrib><creatorcontrib>Chen, Weifeng</creatorcontrib><creatorcontrib>Chen, Gongfa</creatorcontrib><collection>CrossRef</collection><collection>Technology Research Database</collection><collection>Engineering Research Database</collection><collection>Civil Engineering Abstracts</collection><jtitle>Tunnelling and underground space technology</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Zhou, Qianqian</au><au>Situ, Zuxiang</au><au>Teng, Shuai</au><au>Liu, Hanlin</au><au>Chen, Weifeng</au><au>Chen, Gongfa</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Automatic sewer defect detection and severity quantification based on pixel-level semantic segmentation</atitle><jtitle>Tunnelling and underground space technology</jtitle><date>2022-05</date><risdate>2022</risdate><volume>123</volume><spage>104403</spage><pages>104403-</pages><artnum>104403</artnum><issn>0886-7798</issn><eissn>1878-4364</eissn><abstract>•Automated pixel-level sewer defects detection and segmentation.•DeepLabv3+ with five types of backbones are compared with FCN, U-Net and SegNet.•Severity quantification by means of geometric assessment of sewer defects.•Multiple types of sewer defects under complex conditions.•DeepLabv3+-Resnet50 achieved highest PA, mIoU and F1-score, and fastest speed. Underground sewer systems are characterized by their large-scale distributions, high coverage densities and complex defect conditions, which put forward higher requirements for system inspection. Conventional manual inspection is labor-intensive, error-prone and cost inefficient. This paper presents a novel DeepLabv3+-based sewer defect detection and severity quantification method to facilitate automated pixel-level segmentation of sewer defects, from which the defect types, locations, geometric properties and severity levels can be assessed. The effects of different backbone networks on DeepLabv3+ detection accuracy and processing speed were investigated. Three state-of-the-art segmentation methods (SegNet, FCN and U-Net) were further compared to confirm the feasibility of the proposed method. Our results showed that the DeepLabv3+ model, especially with the backbone network Resnet-50, was superior in segmenting multiple types of sewer defects under complex conditions. The obtained PA, mIoU, fwIoU and F1-score were 0.9, 0.53, 0.84 and 0.55, respectively. For individual types of defects, the model worked best in identifying the type of residential walls, followed by disjoints and tree roots. In terms of severity quantification, it was shown that 70% of model predictions were consistent with the ground truths, whereas 30% of predictions likely overestimated the severity levels. The proposed method provides decision-making basis for more accurate and effective sewer inspection and maintenance.</abstract><cop>Oxford</cop><pub>Elsevier Ltd</pub><doi>10.1016/j.tust.2022.104403</doi></addata></record>
fulltext fulltext
identifier ISSN: 0886-7798
ispartof Tunnelling and underground space technology, 2022-05, Vol.123, p.104403, Article 104403
issn 0886-7798
1878-4364
language eng
recordid cdi_proquest_journals_2647396586
source Access via ScienceDirect (Elsevier)
subjects Computer networks
Decision making
DeepLabv3
Defects
Inspection
Pixel level
Pixels
Repair & maintenance
Semantic segmentation
Sensors
Severity quantification
Sewer defect detection
Sewer maintenance
Sewer systems
title Automatic sewer defect detection and severity quantification based on pixel-level semantic segmentation
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T14%3A24%3A11IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Automatic%20sewer%20defect%20detection%20and%20severity%20quantification%20based%20on%20pixel-level%20semantic%20segmentation&rft.jtitle=Tunnelling%20and%20underground%20space%20technology&rft.au=Zhou,%20Qianqian&rft.date=2022-05&rft.volume=123&rft.spage=104403&rft.pages=104403-&rft.artnum=104403&rft.issn=0886-7798&rft.eissn=1878-4364&rft_id=info:doi/10.1016/j.tust.2022.104403&rft_dat=%3Cproquest_cross%3E2647396586%3C/proquest_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2647396586&rft_id=info:pmid/&rft_els_id=S0886779822000438&rfr_iscdi=true