Efficient Lightweight UAV Image Target Detection Model Based on YOLOv7

Efficient Lightweight UAV Image Target Detection Model Based on YOLOv7 | 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 Efficient Lightweight UAV Image Target Detection Model Based on YOLOv7 shuai yin, xiaobo sun This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4084155/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 In the realm of UAV aerial imagery target recognition, detecting small objects presents substantial challenges due to their diminutive size, occlusions among targets, scarcity of distinctive features, and the complexity of backgrounds. To address these issues, this study introduces YOLOv7-PBLight, an innovative, high-precision, and efficient model designed to enhance the accuracy of small object detection in UAV imagery while balancing operational efficiency. Building on the YOLOv7 algorithm, our approach incorporates several key improvements to effectively tackle these challenges: By analyzing the roles of different detection heads, the model eliminates the second subsampling layer and the deepest detection head, thus reducing its receptive field to preserve fine-grained feature information critical for accurate detection of small objects. Furthermore, a Dual-Level Routing Attention (BRA) mechanism is integrated to augment the network's focus on small targets, thereby improving overall algorithm performance. A novel loss function that combines Normalized Gaussian Wasserstein Distance (NWD) with CIOU is introduced to diminish the sensitivity of small objects to positional inaccuracies. Finally, the YOLOv7 backbone is refined using the lightweight network ShuffleNetv2, reducing the model's parameter size and computational burden. Additionally, the utilization of deep separable convolution technology enhances the ELAN and MPConv modules, culminating in the development of YOLOv7-PBiKlight, a lightweight network that achieves a balance between precision and speed. Empirical validation demonstrates that the proposed method effectively resolves the challenges of aerial target detection, significantly improving detection precision with a 4.6% increase in mAP_0.5on the VisDrone2021, reducing the model's parameter size by 67%, and achieving competitive accuracy and speed compared to other algorithms in the field. UAV small target detection YOLOv7 attention mechanism Biformer Model light weighting 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-4084155","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":279254739,"identity":"900ddc10-47de-4be9-8c37-09442d1571d5","order_by":0,"name":"shuai yin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/klEQVRIie2PMUsDMRTHEwLPJbRrjqucHyFQuEn0qxwcZAqCCFJxqFCIm3MnP0bmHA/qUr31oA6tgnM7CHURc06dLh0LzW8I7w//H++FkEjkIOkvcTv6HZsMq6WPvBdWOEUyd/RJQCnbCHsoDKlx9FnwXLQ5qMj61eG1eWeQGnX3rS8GQNjqo+lSmqsCp29fAAOcLU5t6Q+D4VB3KkQ6fst8U6lFYtuBQ9qp1M4rwAQQnd8kdryH4rREblCC0DndWAwrSaOKajpXBfBZmVL74jcG/tKrEdfr0XmRPU6qzY+9v+yfTFafXcqZ2wmM/78d9ZbsYSfQbaAdiUQix8kfXp1O2pLgVX0AAAAASUVORK5CYII=","orcid":"","institution":"Harbin University of Science and Technology","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"shuai","middleName":"","lastName":"yin","suffix":""},{"id":279254740,"identity":"86defa35-9111-4352-a663-a8a1fc59d658","order_by":1,"name":"xiaobo sun","email":"","orcid":"","institution":"Harbin University of Science and Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"xiaobo","middleName":"","lastName":"sun","suffix":""}],"badges":[],"createdAt":"2024-03-12 12:21:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4084155/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4084155/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":53093747,"identity":"8d47989e-3146-4243-8d79-68547fb5c11c","added_by":"auto","created_at":"2024-03-20 13:28:43","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2182506,"visible":true,"origin":"","legend":"","description":"","filename":"515957158.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4084155/v1_covered_a832ee78-aa37-47bf-a006-71f695e09c34.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Efficient Lightweight UAV Image Target Detection Model Based on YOLOv7","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":"UAV small target detection, YOLOv7, attention mechanism, Biformer, Model light weighting","lastPublishedDoi":"10.21203/rs.3.rs-4084155/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4084155/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn the realm of UAV aerial imagery target recognition, detecting small objects presents substantial challenges due to their diminutive size, occlusions among targets, scarcity of distinctive features, and the complexity of backgrounds. To address these issues, this study introduces YOLOv7-PBLight, an innovative, high-precision, and efficient model designed to enhance the accuracy of small object detection in UAV imagery while balancing operational efficiency. Building on the YOLOv7 algorithm, our approach incorporates several key improvements to effectively tackle these challenges: By analyzing the roles of different detection heads, the model eliminates the second subsampling layer and the deepest detection head, thus reducing its receptive field to preserve fine-grained feature information critical for accurate detection of small objects. Furthermore, a Dual-Level Routing Attention (BRA) mechanism is integrated to augment the network's focus on small targets, thereby improving overall algorithm performance. A novel loss function that combines Normalized Gaussian Wasserstein Distance (NWD) with CIOU is introduced to diminish the sensitivity of small objects to positional inaccuracies. Finally, the YOLOv7 backbone is refined using the lightweight network ShuffleNetv2, reducing the model's parameter size and computational burden. Additionally, the utilization of deep separable convolution technology enhances the ELAN and MPConv modules, culminating in the development of YOLOv7-PBiKlight, a lightweight network that achieves a balance between precision and speed. Empirical validation demonstrates that the proposed method effectively resolves the challenges of aerial target detection, significantly improving detection precision with a 4.6% increase in mAP_0.5on the VisDrone2021, reducing the model's parameter size by 67%, and achieving competitive accuracy and speed compared to other algorithms in the field.\u003c/p\u003e","manuscriptTitle":"Efficient Lightweight UAV Image Target Detection Model Based on YOLOv7","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-03-15 03:27:52","doi":"10.21203/rs.3.rs-4084155/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":"f7dc02ca-0563-499b-b4a6-d0d01df5b454","owner":[],"postedDate":"March 15th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-03-20T13:20:27+00:00","versionOfRecord":[],"versionCreatedAt":"2024-03-15 03:27:52","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4084155","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4084155","identity":"rs-4084155","version":["v1"]},"buildId":"-HB7Z8yhvgn0wM9Nzuekk","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}