Efficient Feature Extraction and Fusion for Lightweight Semantic Segmentation Networks

Efficient Feature Extraction and Fusion for Lightweight Semantic Segmentation Networks | 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 Article Efficient Feature Extraction and Fusion for Lightweight Semantic Segmentation Networks Pengfei Yuan, Yuanyuan Wu, Yuan Zeng, Jing Liang, Wu Zeng This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4926441/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 Semantic segmentation, pivotal in applications like autonomous driving and robotics, requires accurate pixel-wise labeling. Here we propose a novel Feature Extraction and Fusion (FEF) module, integrating dilated convolution and depth-wise separable convolution, to swiftly extract multi-scale features with enhanced accuracy and computational efficiency. Our lightweight network demonstrates a 72.6% Mean Intersection over Union (mIoU) on the Cityscapes dataset and achieves an impressive 93.7 FPS on a single GTX 1080Ti GPU, showcasing a competitive balance between speed and accuracy. This work contributes to the ongoing advancements in real-time semantic segmentation by offering a streamlined approach that maintains high performance with minimal computational overhead. The code can be found in https://figshare.com/s/ 5f3ad04a8ba0128b633f Semantic segmentation feature extraction and fusion dilated convolution depth-wise separable convolution 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-4926441","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":359481835,"identity":"56e701e3-6786-47e6-8d57-004799d072db","order_by":0,"name":"Pengfei Yuan","email":"","orcid":"","institution":"ChengDu University of Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Pengfei","middleName":"","lastName":"Yuan","suffix":""},{"id":359481836,"identity":"f7d55fc6-0021-4dbf-87d1-15d5838a5c36","order_by":1,"name":"Yuanyuan Wu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4UlEQVRIiWNgGAWjYHAD5gMMPGBGAtFa2BKAWgxI0sJjQJwWgxs5ZhI/dxxO7J/d8/HD27Y/DPzsOQYMP3fg1yLZe+Zw4ow7ZzdLzm0zYJDseWPA2HsGtxYzkC28bYcTG27kbmPmBWoBGmLAzNiGX4vkX6CW+TdynoG12BOjRRpky4YbOWwQWyQIaLE/86zYWrYt3XjjjTRjyTnnjHkkzjwrONiLR4tke/LGm2/brGXn3Uh++OFNmZwcP1DkwU88WhgYOEBR0QznghPAAXwaGBjYHwCJOvxqRsEoGAWjYGQDALClUwLrwqrWAAAAAElFTkSuQmCC","orcid":"","institution":"ChengDu University of Technology","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Yuanyuan","middleName":"","lastName":"Wu","suffix":""},{"id":359481839,"identity":"2023851b-4168-4fc8-99b5-d97b604a179b","order_by":2,"name":"Yuan Zeng","email":"","orcid":"","institution":"ChengDu University of Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yuan","middleName":"","lastName":"Zeng","suffix":""},{"id":359481842,"identity":"bd30d54f-db41-4548-ab15-cc908b3d9102","order_by":3,"name":"Jing Liang","email":"","orcid":"","institution":"ChengDu University of Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Liang","suffix":""},{"id":359481843,"identity":"de1fb7b5-0ef6-41bd-b95c-4035ce6dd70d","order_by":4,"name":"Wu Zeng","email":"","orcid":"","institution":"ChengDu University of Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Wu","middleName":"","lastName":"Zeng","suffix":""}],"badges":[],"createdAt":"2024-08-16 17:18:03","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4926441/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4926441/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":83333301,"identity":"4b0f33d9-8b62-4575-bc86-085704d78624","added_by":"auto","created_at":"2025-05-23 08:31:46","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2640060,"visible":true,"origin":"","legend":"","description":"","filename":"EfficientFeatureExtractionandFusionforLightweightSemanticSegmentationNetworks.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4926441/v1_covered_2b4368a3-e10f-4ee2-a532-91c71b3a299f.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Efficient Feature Extraction and Fusion for Lightweight Semantic Segmentation Networks","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":"Semantic segmentation, feature extraction and fusion, dilated convolution, depth-wise separable convolution","lastPublishedDoi":"10.21203/rs.3.rs-4926441/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4926441/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Semantic segmentation, pivotal in applications like autonomous driving and robotics, requires accurate pixel-wise labeling. Here we propose a novel Feature Extraction and Fusion (FEF) module, integrating dilated convolution and depth-wise separable convolution, to swiftly extract multi-scale features with enhanced accuracy and computational efficiency. Our lightweight network demonstrates a 72.6% Mean Intersection over Union (mIoU) on the Cityscapes dataset and achieves an impressive 93.7 FPS on a single GTX 1080Ti GPU, showcasing a competitive balance between speed and accuracy. This work contributes to the ongoing advancements in real-time semantic segmentation by offering a streamlined approach that maintains high performance with minimal computational overhead. The code can be found in https://figshare.com/s/ 5f3ad04a8ba0128b633f","manuscriptTitle":"Efficient Feature Extraction and Fusion for Lightweight Semantic Segmentation Networks","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-10-01 18:25:28","doi":"10.21203/rs.3.rs-4926441/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":"448479ff-7cba-4976-9842-ef2819fa6446","owner":[],"postedDate":"October 1st, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-05-23T08:23:38+00:00","versionOfRecord":[],"versionCreatedAt":"2024-10-01 18:25:28","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4926441","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4926441","identity":"rs-4926441","version":["v1"]},"buildId":"cBFmMYwuxLRRLfASyISRj","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}