SAC-YOLO: Efficient Multi-Scale Feature Fusion for Transmission Line Defect Detection

SAC-YOLO: Efficient Multi-Scale Feature Fusion for Transmission Line Defect Detection | 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 SAC-YOLO: Efficient Multi-Scale Feature Fusion for Transmission Line Defect Detection Haotian Yin, Fanghua Liu, Jiankang Yuan, Juntao Fan, Chaojie Xu, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9341141/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Transmission line inspection requires accurate and real-time detection of small defects in complex environments. However, targets such as broken lines, broken insulators, stained insulators, and nests are often small and easily affected by background interference. To address these challenges, this paper proposes an improved YOLOv11n-based detection method. First, the SPPF module is replaced with the AIFI module, which we introduce to enhance intra-scale feature interaction and global contextual modeling on high-level semantic features, thereby improving the representation of small defect targets with limited computational overhead. Second, a C3K2-CFBlock module is constructed by reformulating the feature extraction structure of the original C3K2, enabling efficient integration of local convolutional features and global contextual information for improved long-range dependency modeling with low computational cost. In addition, a Semantic-Guided Multi-Scale Fusion Module (SGMF) is proposed to address the inherent limitations of conventional FPN-based feature aggregation. Instead of relying on direct concatenation or element-wise addition, SGMF introduces channel-aware reweighting and a bidirectional guided fusion mechanism to establish explicit semantic interaction between feature levels, significantly improving the detection performance of small defect targets. Experimental results show that the proposed method improves Recall by 3.5%, and [email protected] :0.95 by 3.8% compared with YOLOv11n, the model achieves real-time inference at 277 FPS, which is comparable to the baseline performance, while maintaining a lightweight structure suitable for real-time deployment on edge devices. Transmission line inspection YOLOv11 Multi-scale Feature Fusion Deep Learning Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-9341141","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":624081175,"identity":"192cebb9-401b-4d11-96bc-77c564202945","order_by":0,"name":"Haotian Yin","email":"","orcid":"","institution":"Jiangsu University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Haotian","middleName":"","lastName":"Yin","suffix":""},{"id":624081176,"identity":"70bfe589-7095-4a84-8174-41bf22f5f4f7","order_by":1,"name":"Fanghua Liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9UlEQVRIiWNgGAWjYBACPmYGhg9AOoGBgfnAgQ8VEnLyhLSwMTMwzoBoYUs8OOOMhbFhAyEtDHAtPMaHedsqEhkOENLCzmPY8HNHbZ7B8QMGB3jnSSQwNjA/fHQDr8N4DBt7zxwvNjiTkHBAcptEHjsDm7FxDn4t5g94244lbjiQcOCA4TaJYsYGHjZpAloMG/+CtJx/2HAgcY5EYsMBIrQ087bVJG64kcxw4GADUVrYCptl2w4kzrzxjOFgwzEJY8NmAn7h5z+8sfFtW11i3/n8z5//1NTJybM3P3yMTwsUHEZiMxNWDgJ1xCkbBaNgFIyCkQkAyIJSrzjPB6EAAAAASUVORK5CYII=","orcid":"","institution":"Jiangsu University of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"Fanghua","middleName":"","lastName":"Liu","suffix":""},{"id":624081177,"identity":"ca2cddfd-55be-4c57-81be-259e939afda4","order_by":2,"name":"Jiankang Yuan","email":"","orcid":"","institution":"Jiangsu University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Jiankang","middleName":"","lastName":"Yuan","suffix":""},{"id":624081178,"identity":"9385a1fe-fa6d-4b4b-8bf1-0947181bc59c","order_by":3,"name":"Juntao Fan","email":"","orcid":"","institution":"Jiangsu University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Juntao","middleName":"","lastName":"Fan","suffix":""},{"id":624081180,"identity":"903b2ee0-b536-44ac-9ca3-df7198e9260c","order_by":4,"name":"Chaojie Xu","email":"","orcid":"","institution":"Jiangsu University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Chaojie","middleName":"","lastName":"Xu","suffix":""},{"id":624081182,"identity":"5acbc689-423a-42bc-b98b-a1f0c708ea89","order_by":5,"name":"Ruibo Tan","email":"","orcid":"","institution":"Jiangsu University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Ruibo","middleName":"","lastName":"Tan","suffix":""}],"badges":[],"createdAt":"2026-04-07 07:26:47","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9341141/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9341141/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108490896,"identity":"2aa23234-65b8-423c-9a3a-1cf6f276dfb9","added_by":"auto","created_at":"2026-05-05 09:49:37","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2655653,"visible":true,"origin":"","legend":"","description":"","filename":"SACYOLOEfficientMultiScaleFeatureFusionforTransmissionLineDefectDetection.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9341141/v1_covered_bc8d9da0-8759-4a72-9fab-9861404ead14.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"SAC-YOLO: Efficient Multi-Scale Feature Fusion for Transmission Line Defect Detection","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Transmission line inspection, YOLOv11, Multi-scale Feature Fusion, Deep Learning","lastPublishedDoi":"10.21203/rs.3.rs-9341141/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9341141/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Transmission line inspection requires accurate and real-time detection of small defects in complex environments. However, targets such as broken lines, broken insulators, stained insulators, and nests are often small and easily affected by background interference. To address these challenges, this paper proposes an improved YOLOv11n-based detection method. First, the SPPF module is replaced with the AIFI module, which we introduce to enhance intra-scale feature interaction and global contextual modeling on high-level semantic features, thereby improving the representation of small defect targets with limited computational overhead. Second, a C3K2-CFBlock module is constructed by reformulating the feature extraction structure of the original C3K2, enabling efficient integration of local convolutional features and global contextual information for improved long-range dependency modeling with low computational cost. In addition, a Semantic-Guided Multi-Scale Fusion Module (SGMF) is proposed to address the inherent limitations of conventional FPN-based feature aggregation. Instead of relying on direct concatenation or element-wise addition, SGMF introduces channel-aware reweighting and a bidirectional guided fusion mechanism to establish explicit semantic interaction between feature levels, significantly improving the detection performance of small defect targets. Experimental results show that the proposed method improves Recall by 3.5%, and [email protected]:0.95 by 3.8% compared with YOLOv11n, the model achieves real-time inference at 277 FPS, which is comparable to the baseline performance, while maintaining a lightweight structure suitable for real-time deployment on edge devices.","manuscriptTitle":"SAC-YOLO: Efficient Multi-Scale Feature Fusion for Transmission Line Defect Detection","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-17 17:12:22","doi":"10.21203/rs.3.rs-9341141/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7e1a8e59-4033-41d6-afa5-9df3645e703a","owner":[],"postedDate":"April 17th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-04-29T18:10:19+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-29T18:24:49+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-17 17:12:22","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9341141","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9341141","identity":"rs-9341141","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}