DGPL-YOLO: Direction-Guided Progressive Learning for Small Fitting Detection in Pressure Pipeline NDT Drawings

DGPL-YOLO: Direction-Guided Progressive Learning for Small Fitting Detection in Pressure Pipeline NDT Drawings | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article DGPL-YOLO: Direction-Guided Progressive Learning for Small Fitting Detection in Pressure Pipeline NDT Drawings Guangheng Li, Chongfan Lyu, Dongyu Li This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9442990/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract Identifying critical fittings in pressure pipelines is essential for planning non-destructive testing (NDT) tasks. However, this process has traditionally been performed manually due to several challenges: small target sizes, varying orientations, complex backgrounds, and limited training data. To address these issues, we propose a detection method based on YOLOv8 with a direction-enhanced attention mechanism. A Direction-Guided Progressive Learning (DGPL) module is introduced to explicitly model the orientation variations of pipe fittings. The method further integrates the MoonNet multi-scale attention module and the PIoU loss function, forming an optimal combination denoted as MoonNet+PIoU+DGPL. Through ablation experiments and comparisons with state-of-the-art (SOTA) methods on a self-built dataset, the method's effectiveness is validated. Through Grad-CAM++ heatmap visualization analysis, this paper further reveals the limitations of traditional ARC metrics in cases of missed and misclassified detections, and proposes the PW-ARC-F1 evaluation metric tailored for engineering applications. Experimental results show that the proposed method achieves an mAP50 of 87.95%, a 9.3% improvement over the baseline, and performs best under the PW-ARC-F1, demonstrating its engineering value. object detection engineering drawings attention mechanism directional enhancement YOLOv8 Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 02 May, 2026 Reviewers agreed at journal 27 Apr, 2026 Reviewers agreed at journal 21 Apr, 2026 Reviewers invited by journal 21 Apr, 2026 Editor assigned by journal 18 Apr, 2026 Submission checks completed at journal 18 Apr, 2026 First submitted to journal 16 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9442990","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":630400996,"identity":"16b252d7-78c1-42e7-9b0e-dfa922de186c","order_by":0,"name":"Guangheng Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1ElEQVRIiWNgGAWjYHACNmYGAxs5+/YGINvAglgtBWnGBjwHQFokiNXy4XDiBokEEIcILbozcsweFxgcTtwu+fzqhh8FEgz87d0JeLWYnTmWbjzDIN145+ycsps9QIdJnDm7Ab+W483HpHkMrGUbbuek3eABajGQyCWg5TBjG1ALM2PDzTNpN/8QpQVii7Pihhvsx24TZwvILzwGacaSPTlst2UMJHgI++UGMMR4/tjI8bMff3bzDYjR3otfCxLgMQCTxCoHAfYHpKgeBaNgFIyCEQQADilHnCdnrNcAAAAASUVORK5CYII=","orcid":"","institution":"Guangxi Financing Guarantee Group Co.,Ltd","correspondingAuthor":true,"prefix":"","firstName":"Guangheng","middleName":"","lastName":"Li","suffix":""},{"id":630400997,"identity":"6a3230df-bf3f-4a7b-ae2e-cd736267dc54","order_by":1,"name":"Chongfan Lyu","email":"","orcid":"","institution":"Guangzhou Academy of Special Equipment Inspection and Testing","correspondingAuthor":false,"prefix":"","firstName":"Chongfan","middleName":"","lastName":"Lyu","suffix":""},{"id":630400998,"identity":"7f88addb-b815-44b2-9816-cd52d056bec3","order_by":2,"name":"Dongyu Li","email":"","orcid":"","institution":"China Construction First Bureau (Group) Co., Ltd. Shenzhen Branch","correspondingAuthor":false,"prefix":"","firstName":"Dongyu","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2026-04-17 02:09:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9442990/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9442990/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108491998,"identity":"f5ede1cd-5c18-4bcd-b50d-85a68b8d6f07","added_by":"auto","created_at":"2026-05-05 09:56:34","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4983562,"visible":true,"origin":"","legend":"","description":"","filename":"DownloadthejournalarticletemplatepackageDecember2024version.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9442990/v1_covered_8d3bd836-2047-4371-806c-b0e24c20db1d.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"DGPL-YOLO: Direction-Guided Progressive Learning for Small Fitting Detection in Pressure Pipeline NDT Drawings","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"signal-image-and-video-processing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"sivp","sideBox":"Learn more about [Signal, Image and Video Processing](http://link.springer.com/journal/11760)","snPcode":"11760","submissionUrl":"https://submission.nature.com/new-submission/11760/3","title":"Signal, Image and Video Processing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"object detection, engineering drawings, attention mechanism, directional enhancement, YOLOv8","lastPublishedDoi":"10.21203/rs.3.rs-9442990/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9442990/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIdentifying critical fittings in pressure pipelines is essential for planning non-destructive testing (NDT) tasks. However, this process has traditionally been performed manually due to several challenges: small target sizes, varying orientations, complex backgrounds, and limited training data. To address these issues, we propose a detection method based on YOLOv8 with a direction-enhanced attention mechanism. A Direction-Guided Progressive Learning (DGPL) module is introduced to explicitly model the orientation variations of pipe fittings. The method further integrates the MoonNet multi-scale attention module and the PIoU loss function, forming an optimal combination denoted as MoonNet+PIoU+DGPL. Through ablation experiments and comparisons with state-of-the-art (SOTA) methods on a self-built dataset, the method's effectiveness is validated. Through Grad-CAM++ heatmap visualization analysis, this paper further reveals the limitations of traditional ARC metrics in cases of missed and misclassified detections, and proposes the PW-ARC-F1 evaluation metric tailored for engineering applications. Experimental results show that the proposed method achieves an mAP50 of 87.95%, a 9.3% improvement over the baseline, and performs best under the PW-ARC-F1, demonstrating its engineering value.\u003c/p\u003e","manuscriptTitle":"DGPL-YOLO: Direction-Guided Progressive Learning for Small Fitting Detection in Pressure Pipeline NDT Drawings","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-30 14:51:04","doi":"10.21203/rs.3.rs-9442990/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-02T23:26:54+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"94519675689546703793514067820551107738","date":"2026-04-27T14:06:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"176772736371057310879005758459950664722","date":"2026-04-21T23:35:29+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-21T22:26:33+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-18T07:18:58+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-18T07:18:53+00:00","index":"","fulltext":""},{"type":"submitted","content":"Signal, Image and Video Processing","date":"2026-04-17T01:58:26+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"signal-image-and-video-processing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"sivp","sideBox":"Learn more about [Signal, Image and Video Processing](http://link.springer.com/journal/11760)","snPcode":"11760","submissionUrl":"https://submission.nature.com/new-submission/11760/3","title":"Signal, Image and Video Processing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"9c73326c-9d98-48e9-a39b-9ffb5dd55d14","owner":[],"postedDate":"April 30th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-02T23:26:54+00:00","index":19,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-30T14:51:05+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-30 14:51:04","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9442990","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9442990","identity":"rs-9442990","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}