Marine garbage identification algorithm based on MGYOLOv7-Tiny network

Marine garbage identification algorithm based on MGYOLOv7-Tiny network | 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 Marine garbage identification algorithm based on MGYOLOv7-Tiny network Guanfang ZUO, Sirui Gu, Xiaotian REN, Changliang Hao, Yiyang Ruan, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5351148/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 Aiming at the problem of marine garbage types detection on YOLOv7-Tiny network, a 2-IDetect network named MGYOLOv7-Tiny is proposed to relieve the contradiction between parameter quantity and detection accuracy. Firstly, the number of detection heads is reduced to balance the demands of accuracy improvement and network slimming. Secondly, local feature information is exchanged by light-weight convolution named GSConv. Model parameter quantity is reduced by the training-time and inference-time illating with different Re-parameterization Visual Geometry Group (RepVGG) structures. The activation function of Global Attention Mechanism (GAM) is changed to Hard Swish, so the problem of parameter update near the origin is completely solved. The pixel continuity of upper sampling feature maps is improved by the use of Content-Aware ReAssembly of Features (CARAFE). The prior box size conforming to small marine debris is obtained by using the Kmeans++ algorithm. Finally, Alpha-IoU is used to improve the bounding box regression accuracy of small samples by adjusting the power parameter of . Experimental results show that MGYOLOv7-Tiny has the detection precision of 91.72% and the parameter quantity is 6.7MB on trash\_ICRA19. The universality of MGYOLOv7-Tiny is demonstrated in a small target dataset named Visdrone2019. YOLOv7-Tiny network marine garbage types detection 2- IDetect network Re-parameterization Visual Geometry Group 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-5351148","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":373501956,"identity":"ad0d7d3f-33d6-47ce-8d9d-ca76e0177920","order_by":0,"name":"Guanfang ZUO","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAt0lEQVRIiWNgGAWjYFACxocPoCwDYrUwG8OUEq/FTII0LfL9i9kqfrYdTmxgb94mwVBzh7AWxhmP2W72tqUlNvAcK5NgOPaMCGdJnD92m3GbTWKDRI6ZBGPDYcJa2CQOsxUzbpNIbJB/Q6QWHv5mNmaILTxEapGQYGaW7P2XZtzGk1ZskXCMCC3y/YcZP/w4c1i2n/3wxhsfaojQwiCRAKHZQEQCERoYGPgPEKVsFIyCUTAKRjIAAKC2NH15fRN1AAAAAElFTkSuQmCC","orcid":"","institution":"Wuxi University","correspondingAuthor":true,"prefix":"","firstName":"Guanfang","middleName":"","lastName":"ZUO","suffix":""},{"id":373501957,"identity":"e289e694-4f98-42cf-821d-bbf228bc0175","order_by":1,"name":"Sirui Gu","email":"","orcid":"","institution":"Wuxi University","correspondingAuthor":false,"prefix":"","firstName":"Sirui","middleName":"","lastName":"Gu","suffix":""},{"id":373501958,"identity":"30d27edb-b4dc-47a0-8b15-43356bc38f07","order_by":2,"name":"Xiaotian REN","email":"","orcid":"","institution":"Nanjing University of Information Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Xiaotian","middleName":"","lastName":"REN","suffix":""},{"id":373501959,"identity":"42f4a232-4f8f-468e-a870-7631a2c164d8","order_by":3,"name":"Changliang Hao","email":"","orcid":"","institution":"Wuxi University","correspondingAuthor":false,"prefix":"","firstName":"Changliang","middleName":"","lastName":"Hao","suffix":""},{"id":373501960,"identity":"5e62c9ba-1633-4747-b086-6efde984165a","order_by":4,"name":"Yiyang Ruan","email":"","orcid":"","institution":"Wuxi University","correspondingAuthor":false,"prefix":"","firstName":"Yiyang","middleName":"","lastName":"Ruan","suffix":""},{"id":373501961,"identity":"c219f8f8-abaa-47dc-863c-1ca8d78d7ec9","order_by":5,"name":"Yanyou SHA","email":"","orcid":"","institution":"Nanjing University of Information Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Yanyou","middleName":"","lastName":"SHA","suffix":""}],"badges":[],"createdAt":"2024-10-29 05:08:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5351148/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5351148/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":71805811,"identity":"7d71fd80-80ee-4aee-b60a-5c5fe73fda11","added_by":"auto","created_at":"2024-12-18 17:31:44","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":536882,"visible":true,"origin":"","legend":"","description":"","filename":"MarinegarbageidentificationalgorithmbasedonMGYOLOv7Tinynetwork.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5351148/v1_covered_e09283d4-49e4-4e80-9da7-36f0ea5643f3.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Marine garbage identification algorithm based on MGYOLOv7-Tiny network","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":"YOLOv7-Tiny network, marine garbage, types detection, 2- IDetect network, Re-parameterization Visual Geometry Group","lastPublishedDoi":"10.21203/rs.3.rs-5351148/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5351148/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAiming at the problem of marine garbage types detection on YOLOv7-Tiny network, a 2-IDetect network named MGYOLOv7-Tiny is proposed to relieve the contradiction between parameter quantity and detection accuracy. Firstly, the number of detection heads is reduced to balance the demands of accuracy improvement and network slimming. Secondly, local feature information is exchanged by light-weight convolution named GSConv. Model parameter quantity is reduced by the training-time and inference-time illating with different Re-parameterization Visual Geometry Group (RepVGG) structures. The activation function of Global Attention Mechanism (GAM) is changed to Hard Swish, so the problem of parameter update near the origin is completely solved. The pixel continuity of upper sampling feature maps is improved by the use of Content-Aware ReAssembly of Features (CARAFE). The prior box size conforming to small marine debris is obtained by using the Kmeans++ algorithm. Finally, Alpha-IoU is used to improve the bounding box regression accuracy of small samples by adjusting the power parameter of . Experimental results show that MGYOLOv7-Tiny has the detection precision of 91.72% and the parameter quantity is 6.7MB on trash\\_ICRA19. The universality of MGYOLOv7-Tiny is demonstrated in a small target dataset named Visdrone2019.\u003c/p\u003e","manuscriptTitle":"Marine garbage identification algorithm based on MGYOLOv7-Tiny network","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-11-14 03:07:45","doi":"10.21203/rs.3.rs-5351148/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":"32c49490-81a8-496d-989b-6b5ec92a14ba","owner":[],"postedDate":"November 14th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-12-18T17:23:33+00:00","versionOfRecord":[],"versionCreatedAt":"2024-11-14 03:07:45","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5351148","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5351148","identity":"rs-5351148","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}