No Cover Image

Journal article 185 views 13 downloads

RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning

Yuanyuan Hu Orcid Logo, Ning Chen, Hancheng Zhang, Yue Hou Orcid Logo, Pengfei Liu

Research, Volume: 8

Swansea University Author: Yue Hou Orcid Logo

  • 70722.pdf

    PDF | Version of Record

    Copyright © 2025 Yuanyuan Hu et al. Exclusive licensee Science and Technology Review Publishing House. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution License (CC BY 4.0).

    Download (12.1MB)

Check full text

DOI (Published version): 10.34133/research.0833

Abstract

During the designed service life, road infrastructures will bear repeated loading conditions from vehicle weights and environmental conditions, resulting in the inevitable occurrence of road distresses including cracks, potholes, etc. The traditional inspection methods by transportation engineers ar...

Full description

Published in: Research
ISSN: 2639-5274
Published: American Association for the Advancement of Science (AAAS) 2025
Online Access: Check full text

URI: https://cronfa.swan.ac.uk/Record/cronfa70722
first_indexed 2025-10-18T18:04:42Z
last_indexed 2025-12-05T18:10:25Z
id cronfa70722
recordtype SURis
fullrecord <?xml version="1.0"?><rfc1807><datestamp>2025-12-04T08:52:13.1327337</datestamp><bib-version>v2</bib-version><id>70722</id><entry>2025-10-18</entry><title>RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning</title><swanseaauthors><author><sid>92bf566c65343cb3ee04ad963eacf31b</sid><ORCID>0000-0002-4334-2620</ORCID><firstname>Yue</firstname><surname>Hou</surname><name>Yue Hou</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2025-10-18</date><deptcode>ACEM</deptcode><abstract>During the designed service life, road infrastructures will bear repeated loading conditions from vehicle weights and environmental conditions, resulting in the inevitable occurrence of road distresses including cracks, potholes, etc. The traditional inspection methods by transportation engineers are normally costly and labor-intensive. In recent years, artificial intelligence (AI)-based road distress detection methods have been widely used as convenient and automated approaches, while the AI-based methods heavily depend on a large amount of high-quality images, limiting the real engineering applications. To address the issues, this study introduces RoadDiffBox, a novel framework employing controlled image generation and semi-supervised learning. The framework addresses dataset imbalances through class control and accelerates image generation by utilizing the denoising diffusion implicit model&#x2019;s reverse process sampling method, while employing knowledge distillation techniques optimized for resource-constrained mobile devices. It generates diverse and high-quality road distress images with automatic bounding box annotations, substantially reducing manual labeling requirements. Test results show that RoadDiffBox demonstrates strong generalizability across geographic regions (Germany, China, and India) and shows cross-domain potential in medical imaging applications. Performance evaluations demonstrate RoadDiffBox&#x2019;s effectiveness, with classification models achieving an F1-score of 0.95 and detection models reaching a mean average precision (mAP@50) of 0.95 and an F1-score of 0.91 in controlled settings, while maintaining robust performance (an F1-score of 0.86 and a mAP@50 of 0.91) during on-site testing in real-world conditions. On server-class hardware, the model achieves generation times as low as 0.18 s per image. It is discovered that RoadDiffBox can serve as a scalable and efficient solution for real-time road maintenance with limited datasets.</abstract><type>Journal Article</type><journal>Research</journal><volume>8</volume><journalNumber/><paginationStart/><paginationEnd/><publisher>American Association for the Advancement of Science (AAAS)</publisher><placeOfPublication/><isbnPrint/><isbnElectronic/><issnPrint/><issnElectronic>2639-5274</issnElectronic><keywords/><publishedDay>25</publishedDay><publishedMonth>8</publishedMonth><publishedYear>2025</publishedYear><publishedDate>2025-08-25</publishedDate><doi>10.34133/research.0833</doi><url/><notes/><college>COLLEGE NANME</college><department>Aerospace, Civil, Electrical, and Mechanical Engineering</department><CollegeCode>COLLEGE CODE</CollegeCode><DepartmentCode>ACEM</DepartmentCode><institution>Swansea University</institution><apcterm>Another institution paid the OA fee</apcterm><funders>This paper is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - SFB/TRR 339, Project-ID 453596084</funders><projectreference/><lastEdited>2025-12-04T08:52:13.1327337</lastEdited><Created>2025-10-18T19:01:44.7614288</Created><path><level id="1">Faculty of Science and Engineering</level><level id="2">School of Aerospace, Civil, Electrical, General and Mechanical Engineering - Civil Engineering</level></path><authors><author><firstname>Yuanyuan</firstname><surname>Hu</surname><orcid>0009-0002-3653-4947</orcid><order>1</order></author><author><firstname>Ning</firstname><surname>Chen</surname><order>2</order></author><author><firstname>Hancheng</firstname><surname>Zhang</surname><order>3</order></author><author><firstname>Yue</firstname><surname>Hou</surname><orcid>0000-0002-4334-2620</orcid><order>4</order></author><author><firstname>Pengfei</firstname><surname>Liu</surname><order>5</order></author></authors><documents><document><filename>70722__35387__113d27d6fc024a908322e73dad98564c.pdf</filename><originalFilename>70722.pdf</originalFilename><uploaded>2025-10-18T19:04:36.2823847</uploaded><type>Output</type><contentLength>12692087</contentLength><contentType>application/pdf</contentType><version>Version of Record</version><cronfaStatus>true</cronfaStatus><documentNotes>Copyright &#xA9; 2025 Yuanyuan Hu et al. Exclusive licensee Science and Technology Review Publishing House. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution License (CC BY 4.0).</documentNotes><copyrightCorrect>true</copyrightCorrect><language>eng</language><licence>https://creativecommons.org/licenses/by/4.0/</licence></document></documents><OutputDurs/></rfc1807>
spelling 2025-12-04T08:52:13.1327337 v2 70722 2025-10-18 RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning 92bf566c65343cb3ee04ad963eacf31b 0000-0002-4334-2620 Yue Hou Yue Hou true false 2025-10-18 ACEM During the designed service life, road infrastructures will bear repeated loading conditions from vehicle weights and environmental conditions, resulting in the inevitable occurrence of road distresses including cracks, potholes, etc. The traditional inspection methods by transportation engineers are normally costly and labor-intensive. In recent years, artificial intelligence (AI)-based road distress detection methods have been widely used as convenient and automated approaches, while the AI-based methods heavily depend on a large amount of high-quality images, limiting the real engineering applications. To address the issues, this study introduces RoadDiffBox, a novel framework employing controlled image generation and semi-supervised learning. The framework addresses dataset imbalances through class control and accelerates image generation by utilizing the denoising diffusion implicit model’s reverse process sampling method, while employing knowledge distillation techniques optimized for resource-constrained mobile devices. It generates diverse and high-quality road distress images with automatic bounding box annotations, substantially reducing manual labeling requirements. Test results show that RoadDiffBox demonstrates strong generalizability across geographic regions (Germany, China, and India) and shows cross-domain potential in medical imaging applications. Performance evaluations demonstrate RoadDiffBox’s effectiveness, with classification models achieving an F1-score of 0.95 and detection models reaching a mean average precision (mAP@50) of 0.95 and an F1-score of 0.91 in controlled settings, while maintaining robust performance (an F1-score of 0.86 and a mAP@50 of 0.91) during on-site testing in real-world conditions. On server-class hardware, the model achieves generation times as low as 0.18 s per image. It is discovered that RoadDiffBox can serve as a scalable and efficient solution for real-time road maintenance with limited datasets. Journal Article Research 8 American Association for the Advancement of Science (AAAS) 2639-5274 25 8 2025 2025-08-25 10.34133/research.0833 COLLEGE NANME Aerospace, Civil, Electrical, and Mechanical Engineering COLLEGE CODE ACEM Swansea University Another institution paid the OA fee This paper is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - SFB/TRR 339, Project-ID 453596084 2025-12-04T08:52:13.1327337 2025-10-18T19:01:44.7614288 Faculty of Science and Engineering School of Aerospace, Civil, Electrical, General and Mechanical Engineering - Civil Engineering Yuanyuan Hu 0009-0002-3653-4947 1 Ning Chen 2 Hancheng Zhang 3 Yue Hou 0000-0002-4334-2620 4 Pengfei Liu 5 70722__35387__113d27d6fc024a908322e73dad98564c.pdf 70722.pdf 2025-10-18T19:04:36.2823847 Output 12692087 application/pdf Version of Record true Copyright © 2025 Yuanyuan Hu et al. Exclusive licensee Science and Technology Review Publishing House. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution License (CC BY 4.0). true eng https://creativecommons.org/licenses/by/4.0/
title RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning
spellingShingle RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning
Yue Hou
title_short RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning
title_full RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning
title_fullStr RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning
title_full_unstemmed RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning
title_sort RoadDiffBox: Automatic Road Distress Diagnosis through Controlled Image Generation and Semi-Supervised Learning
author_id_str_mv 92bf566c65343cb3ee04ad963eacf31b
author_id_fullname_str_mv 92bf566c65343cb3ee04ad963eacf31b_***_Yue Hou
author Yue Hou
author2 Yuanyuan Hu
Ning Chen
Hancheng Zhang
Yue Hou
Pengfei Liu
format Journal article
container_title Research
container_volume 8
publishDate 2025
institution Swansea University
issn 2639-5274
doi_str_mv 10.34133/research.0833
publisher American Association for the Advancement of Science (AAAS)
college_str Faculty of Science and Engineering
hierarchytype
hierarchy_top_id facultyofscienceandengineering
hierarchy_top_title Faculty of Science and Engineering
hierarchy_parent_id facultyofscienceandengineering
hierarchy_parent_title Faculty of Science and Engineering
department_str School of Aerospace, Civil, Electrical, General and Mechanical Engineering - Civil Engineering{{{_:::_}}}Faculty of Science and Engineering{{{_:::_}}}School of Aerospace, Civil, Electrical, General and Mechanical Engineering - Civil Engineering
document_store_str 1
active_str 0
description During the designed service life, road infrastructures will bear repeated loading conditions from vehicle weights and environmental conditions, resulting in the inevitable occurrence of road distresses including cracks, potholes, etc. The traditional inspection methods by transportation engineers are normally costly and labor-intensive. In recent years, artificial intelligence (AI)-based road distress detection methods have been widely used as convenient and automated approaches, while the AI-based methods heavily depend on a large amount of high-quality images, limiting the real engineering applications. To address the issues, this study introduces RoadDiffBox, a novel framework employing controlled image generation and semi-supervised learning. The framework addresses dataset imbalances through class control and accelerates image generation by utilizing the denoising diffusion implicit model’s reverse process sampling method, while employing knowledge distillation techniques optimized for resource-constrained mobile devices. It generates diverse and high-quality road distress images with automatic bounding box annotations, substantially reducing manual labeling requirements. Test results show that RoadDiffBox demonstrates strong generalizability across geographic regions (Germany, China, and India) and shows cross-domain potential in medical imaging applications. Performance evaluations demonstrate RoadDiffBox’s effectiveness, with classification models achieving an F1-score of 0.95 and detection models reaching a mean average precision (mAP@50) of 0.95 and an F1-score of 0.91 in controlled settings, while maintaining robust performance (an F1-score of 0.86 and a mAP@50 of 0.91) during on-site testing in real-world conditions. On server-class hardware, the model achieves generation times as low as 0.18 s per image. It is discovered that RoadDiffBox can serve as a scalable and efficient solution for real-time road maintenance with limited datasets.
published_date 2025-08-25T05:31:31Z
_version_ 1851098075499593728
score 11.089386

Similar Items