[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-502070-105":59,"doc-detail-502070-en":131},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","citystar-structured-and-topology-aware-reasoning-for-open-vocabulary-urban-3d-grounding-paper","CitySTAR: Structured and Topology-Aware Reasoning for Open-Vocabulary Urban 3D Grounding - Paper","","3D grounding localizes target entities in complex scenes from natural-language descriptions, yet existing approaches often depend on feature similarity or direct matching, failing to connect language intent with the implicit semantic and geometric structures inside billion-scale urban point clouds. CitySTAR reformulates city-scale 3D grounding as structured constraint reasoning, organizing description semantics into computable cross-modal constraints over open-vocabulary 3D entities, attributes, and spatial relations. It uses training-free graph-based topology reasoning plus reflective cross-modal grounding to improve disambiguation and interpretability. CitySTAR-3D enhances semantic coverage, completeness, fidelity, and relation complexity, and experiments show consistent gains with strong generalization.",{"@graph":69,"@context":123},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/citystar-structured-and-topology-aware-reasoning-for-open-vocabulary-urban-3d-grounding-paper/502070/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/citystar-structured-and-topology-aware-reasoning-for-open-vocabulary-urban-3d-grounding-paper/502070.png","ImageObject",300,407,{"name":92,"@type":93},"นรินทร์","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-15","2026-10-03",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",12,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"What problem does CitySTAR address in city-scale 3D grounding?","Question",{"text":113,"@type":114},"CitySTAR targets the difficulty of grounding open-vocabulary, relational queries in billion-scale urban point clouds, where prior methods rely on matching that cannot reliably verify intended target-context topology.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How does CitySTAR reformulate city-scale 3D grounding?",{"text":118,"@type":114},"It reformulates the task as structured constraint reasoning, organizing description semantics into cross-modal constraints over open-vocabulary 3D entities, attributes, and spatial relations.",{"name":120,"@type":111,"acceptedAnswer":121},"What is CitySTAR-3D and why is it introduced?",{"text":122,"@type":114},"CitySTAR-3D is an enhanced benchmark that improves semantic coverage, instance completeness, bounding-box fidelity, and spatial-relation complexity to better support the city-scale open-vocabulary setting.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},502070,1791097755,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":105,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":144,"read_time":41},2336475104957,"https://ap-avatar.wpscdn.com/avatar/22000c4c6bd8a5076e1?x-image-process=image/resize,m_fixed,w_180,h_180&k=1787554080175789136","arXiv : 2609 . 199 1 1v 1 [cs .CV] 17 Sep 2026  \nCitySTAR: Structured and Topology-Aware Reasoning for Open-Vocabulary Urban 3D Grounding  \nShuai Zhang 1 ,† Hongye Hou 1 ,† Qinghe Liu 1 Zhuoxiao Li 1 Dongli Wu 1 Jing Ou 1 Yuan Liu2 Wufan Zhao 1 ,∗  \n1 HKUST(GZ) , 2 HKUST  \nAbstract  \n3D grounding aims to localize target entities in complex scenes from natural language and plays a fundamental role in embodied perception and spatial reasoning. However, existing approaches mostly rely on feature similarity or direct matching, making it difficult to connect naturallanguage intent with the implicit semantic and geometric structures hidden in billion-scale urban point clouds. We reformulate city-scale 3D grounding as structured constraint reasoning, where description semantics are organized into computable cross-modal constraints over openvocabulary 3D entities, attributes, and spatial relations. We present CitySTAR, a training-free framework for reasoning-driven urban 3D grounding. CitySTAR lifts raw billion-scale urban point clouds into a query-ready scene graph of open-vocabulary 3D instances, with CodeLLMdriven tools supplying multimodal evidence for node attributes and 3D spatial relations. It then models target-context topology with paired hypergraphs and performs bidirectional topology verification for structural disambiguation. Finally, a Reflective Cross-modal Grounding module integrates topology consistency and candidate-centered 2D visual evidence to make decisions over a metric-aware 3D context graph. To further support this setting, we introduce CitySTAR-3D, an enhanced benchmark that improves semantic coverage, instance completeness, bounding-box fidelity, and spatial-relation complexity in city-scale 3D grounding. Extensive experiments show that CitySTAR consistently improves open-world urban 3D grounding while maintaining strong interpretability and generalization.  \nFigure 1 CitySTAR upgrades city-scale 3D grounding from candidate matching to structured reasoning. It combines actionable cue organization, query-ready scene graphs, and metric-aware topology-aware grounding to deliver better target disambiguation and more accurate localization.  \n1 Introduction  \n3D visual grounding aims to localize target entities in a 3D scene according to natural-language descriptions, serving as a key interface between language understanding and spatial perception [3] . While this task has been widely studied in indoor environments, extending it to city-scale scenes introduces a substantially different problem. Urban point clouds often contain billion-scale points and repeated object patterns. Incomplete observations and dense similar-looking entities further make city-scale grounding a relational reasoning problem rather than simple object retrieval. A query such as “the white flat-roof warehouse behind the tree cluster and to the left of the road” requires the model to identify the target category, recognize visual attributes, locate contextual objects, and verify spatial topology in a large urban context. This makes city-scale 3D grounding important for urban digital twins, embodied navigation, and large-scale spatial reasoning [5, 22 , 29 , 37] .  \nExisting grounding methods have made steady progress in 2D vision [2, 10 , 34], indoor 3D scenes [15, 20 , 23 , 24], and recent city-scale settings [18, 25] . However, most methods still follow a matching-centered paradigm, where language is aligned with object features, candidate regions, or learned embeddings. CityRefer [25] selects targets from a predefined candidate set through joint embedding and similarity matching. CityAnchor [18] improves the search process with a coarse-to-fine localization strategy. Their semantic spaces remain constrained by predefined category sets, where CityRefer mainly covers four target categories and CityAnchor extends the scope to nine categories. This closed-set design restricts grounding to a limited portion of urban entities and makes it difficul","cbCaii4zb8Yw14mw","https://ap.wps.com/l/cbCaii4zb8Yw14mw","pdf",25260100,"English","# Abstract\n# 1 Introduction","[{\"question\":\"What problem does CitySTAR address in city-scale 3D grounding?\",\"answer\":\"CitySTAR targets the difficulty of grounding open-vocabulary, relational queries in billion-scale urban point clouds, where prior methods rely on matching that cannot reliably verify intended target-context topology.\"},{\"question\":\"How does CitySTAR reformulate city-scale 3D grounding?\",\"answer\":\"It reformulates the task as structured constraint reasoning, organizing description semantics into cross-modal constraints over open-vocabulary 3D entities, attributes, and spatial relations.\"},{\"question\":\"What is CitySTAR-3D and why is it introduced?\",\"answer\":\"CitySTAR-3D is an enhanced benchmark that improves semantic coverage, instance completeness, bounding-box fidelity, and spatial-relation complexity to better support the city-scale open-vocabulary setting.\"}]","CitySTAR: Structured and Topology-Aware Reasoning for Open-Vocabulary Urban 3D Grounding - Paper | PDF",1791028490]