[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-327990-105":3,"doc-detail-327990-en":81,"detail-sidebar-cat-0-en-105":99},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","targeted-structure-completion-for-sparse-view-3d-reconstruction-in-autonomous-driving-focusgs","Targeted Structure Completion for Sparse-View 3D Reconstruction in Autonomous Driving - FocusGS","","Reconstructing 3D scene structures from sparse, low-overlap observations remains a core challenge in autonomous driving, where limited camera view overlap reduces geometric consistency and degrades reconstruction quality. Recent voxel-based 3D Gaussian approaches improve robustness but waste computation by uniformly processing the entire 3D volume with many redundant Gaussians. FocusGS targets efficiency by decoupling structural completion from deterministic regions and concentrating computation on ambiguous geometry, localized via a 3D Geometric Ambiguity Manifold. A lightweight module then performs targeted Gaussian queries in the sparse topological subspace, achieving improved efficiency–quality trade-offs, ~74% fewer Gaussians, and ~34% lower rendering time on driving-centric benchmarks.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/targeted-structure-completion-for-sparse-view-3d-reconstruction-in-autonomous-driving-focusgs/327990/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/targeted-structure-completion-for-sparse-view-3d-reconstruction-in-autonomous-driving-focusgs/327990.png","ImageObject",300,407,{"name":42,"@type":43},"Rowan","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-11-05","2026-09-21",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",13,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem does FocusGS address in sparse-view 3D reconstruction for driving?","Question",{"text":63,"@type":64},"FocusGS targets the reduced reconstruction quality caused by sparse, low-overlap observations in autonomous driving, while avoiding the computational redundancy of uniform voxel-based processing.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How does FocusGS improve efficiency compared with voxel-based Gaussian methods?",{"text":68,"@type":64},"It shifts from global densification to targeted structural completion, concentrating computations only on regions with geometric ambiguity instead of instantiating and optimizing Gaussians across the whole 3D volume.",{"name":70,"@type":61,"acceptedAnswer":71},"How are critical ambiguous regions located and handled in FocusGS?",{"text":72,"@type":64},"FocusGS derives a 3D Geometric Ambiguity Manifold to localize areas prone to occlusion and high uncertainty, then runs a lightweight completion module that selectively instantiates and optimizes continuous Gaussian queries within that sparse subspace.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},327990,1790645701,{"code":4,"msg":82,"data":83},"success",{"doc_id":79,"user_id":84,"nickname":42,"user_avatar":85,"doc_module":4,"category_id":86,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":87,"file_id":88,"file_url":89,"file_type":90,"file_size":91,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":92,"language":93,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":94,"faqs":95,"seo_title":96,"seo_description":12,"update_tm":97,"read_time":98,"preview_page_count":4},1099514067415,"https://ap-avatar.wpscdn.com/avatar/100002539d78ffe74a7?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779092875211072502",8,"arXiv :2607 .0466 1v 1 [ cs .CV] 6 Jul 2026  \nTargeted Structure Completion for Sparse-View 3D Reconstruction in Autonomous Driving  \nGuoqing Wang 1 , Pin Tang 1 , Xiangxuan Ren 1 , Liping Hou2 , and Chao Ma 1 , ⋆  \n1 MoE Key Lab of Artificial Intelligence, Institute of AI,  \nShanghai Jiao Tong University, Shanghai, China  \n2 Central Research Institute, Huawei, Beijing, China  \n{[guoqing.wang](guoqing.wang),pin.tang,bunny_renxiangxuan,[chaoma}@sjtu.edu.cn](chaoma}@sjtu.edu.cn)[ ](chaoma}@sjtu.edu.cn)[houliping1@huawei.com](houliping1@huawei.com)[ ](houliping1@huawei.com)Project page: [https://focusgs.github.io/](https://focusgs.github.io/)  \nAbstract. Reconstructing 3D scene structures from sparse, low-overlap observations remains a fundamental challenge in autonomous driving.  \nRecent state-of-the-art frameworks achieve promising results by incorporating voxel-based Gaussians, but incur substantial computational redundancy due to a uniform volumetric processing strategy. To bridge the gap between the efficiency of pixel-based Gaussian methods and the structural completeness of voxel-based Gaussian approaches, we propose FocusGS, a simple yet effective framework that shifts the paradigm from global densification to targeted structural completion. Our central insight is that structural completion should be decoupled from deterministic regions, with computation concentrated exclusively on areas exhibiting geometric ambiguity. Specifically, FocusGS addresses the localization challenge by deriving a 3D Geometric Ambiguity Manifold to accurately isolate localized areas prone to occlusion and high geometric uncertainty.  \nTo overcome the subsequent manifold completion challenge, we design a lightweight targeted structure completion module that selectively instantiates and optimizes continuous Gaussian queries strictly within this unstructured, sparse topological subspace. Extensive experiments demonstrate that FocusGS achieves a superior efficiency-quality trade-off, advancing state-of-the-art performance on driving-centric benchmarks while naturally reducing the total number of Gaussians by ∼74% and decreasing rendering time by ∼34% .  \nKeywords: 3D Scene Reconstruction · Geometric Ambiguity · Targeted Structure Completion  \n1 Introduction  \nReconstructing 3D scene structures from sparse observations is a fundamental problem in computer vision and graphics [5, 8, 37, 49, 62] . Recent feed-forward frameworks [5, 8], built upon 3D Gaussian Splatting (3DGS) [21], directly predict pixel-based 3D Gaussians from sparse inputs in a single forward pass. These  \n⋆ Corresponding author.  \n2 G. Wang et al.  \nFig. 1: Comparison of feed-forward 3D Gaussian Splatting methods. (a, b) Prior works rely on either pixel- [5, 8] or voxel-based [44] predictors. (c) Omni-Scene [49] combines both but suffers from redundant Gaussians. (d) Our FocusGS introduces targeted structure completion based on pixel-based features, achieving high-fidelity reconstruction with minimal Gaussian overhead.  \napproaches offer remarkable efficiency and generalization, particularly in largescale dynamic environments. However, they implicitly assume substantial overlap between input views to enforce geometric consistency. This assumption confronts the challenge in ego-centric autonomous driving, where cameras are designed for maximal coverage with minimal overlap ( \u003C 15%), leading to degraded reconstruction quality.  \nTo alleviate the reliance on cross-view overlap, voxel-based Gaussian methods [15,44,66] lift 2D image features into 3D space, thereby improving robustness to occlusion and frustum truncation (Fig. 1(b)) . In particular, state-of-the-art Omni-Gaussian frameworks [49] adopt a dual-branch architecture that integrates pixel-and voxel-based Gaussians to infer and complete occluded regions explicitly. Although effective, this design introduces substantial computational overhead due to its uniform volumetric processing strategy (Fig. 1(c)), which instantiates and opt","cbCaipcO1u48oyhF","https://ap.wps.com/l/cbCaipcO1u48oyhF","pdf",1889349,19,"English","# Introduction\n## Problem: sparse views and low overlap\n## Limitation of uniform voxel-based processing\n## Key idea: FocusGS targeted structural completion\n## Localization: 3D Geometric Ambiguity Manifold\n## Completion: lightweight targeted Gaussian queries","[{\"question\":\"What problem does FocusGS address in sparse-view 3D reconstruction for driving?\",\"answer\":\"FocusGS targets the reduced reconstruction quality caused by sparse, low-overlap observations in autonomous driving, while avoiding the computational redundancy of uniform voxel-based processing.\"},{\"question\":\"How does FocusGS improve efficiency compared with voxel-based Gaussian methods?\",\"answer\":\"It shifts from global densification to targeted structural completion, concentrating computations only on regions with geometric ambiguity instead of instantiating and optimizing Gaussians across the whole 3D volume.\"},{\"question\":\"How are critical ambiguous regions located and handled in FocusGS?\",\"answer\":\"FocusGS derives a 3D Geometric Ambiguity Manifold to localize areas prone to occlusion and high uncertainty, then runs a lightweight completion module that selectively instantiates and optimizes continuous Gaussian queries within that sparse subspace.\"}]","Targeted Structure Completion for Sparse-View 3D Reconstruction in Autonomous Driving - FocusGS | PDF",1789978448,48,{"code":4,"msg":82,"data":100},[101,105,109,113,118,123,128,131,136,139,143],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":102,"show_sort_weight":103,"slug":104},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":106,"show_sort_weight":107,"slug":108},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":110,"show_sort_weight":111,"slug":112},"Exam",70,"exam",{"id":114,"doc_module":4,"doc_module_name":25,"category_name":115,"show_sort_weight":116,"slug":117},5,"Comic",60,"comic",{"id":119,"doc_module":4,"doc_module_name":25,"category_name":120,"show_sort_weight":121,"slug":122},6,"Technology",50,"technology",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":126,"slug":127},7,"Healthcare",40,"healthcare",{"id":86,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":129,"slug":130},30,"research-report",{"id":132,"doc_module":4,"doc_module_name":25,"category_name":133,"show_sort_weight":134,"slug":135},9,"Religion & Spirituality",20,"religion-spirituality",{"id":134,"doc_module":4,"doc_module_name":25,"category_name":137,"show_sort_weight":134,"slug":138},"World Cup","world-cup",{"id":140,"doc_module":4,"doc_module_name":25,"category_name":141,"show_sort_weight":140,"slug":142},10,"Lifestyle","lifestyle",{"id":92,"doc_module":4,"doc_module_name":25,"category_name":144,"show_sort_weight":114,"slug":145},"General","general"]