[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-648045-105":59,"doc-detail-648045-en":130},{"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":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","deeppatent-large-scale-patent-drawing-recognition-and-retrieval-paper","DeepPatent - Large-scale Patent Drawing Recognition and Retrieval - Paper","","DeepPatent addresses analysis and retrieval of technical drawings by introducing a large-scale design patent dataset for image retrieval, offering 350,000+ drawings. The dataset enables fine-grained, within-domain retrieval associations without cross-domain supervision. A baseline model, PatentNet, is built following retrieval best practices for static images. Experiments show PatentNet outperforms alternative deep learning methods and classic computer-vision descriptors, establishing patent drawing retrieval as an open, research-driving benchmark.",{"@graph":69,"@context":122},[70,84,105],{"@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/deeppatent-large-scale-patent-drawing-recognition-and-retrieval-paper/648045/",{"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/deeppatent-large-scale-patent-drawing-recognition-and-retrieval-paper/648045.png","ImageObject",300,407,{"name":92,"@type":93},"Liam","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-25","2026-10-23",true,{"@type":102,"interactionType":103,"userInteractionCount":14},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What problem does DeepPatent target?","Question",{"text":112,"@type":113},"DeepPatent focuses on analyzing and retrieving technical drawings, particularly design patent drawings, using visual content rather than text-only methods.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"What distinguishes the DeepPatent dataset from existing datasets?",{"text":117,"@type":113},"DeepPatent provides fine-grained image retrieval associations within the patent drawing collection and does not rely on cross-domain supervision.",{"name":119,"@type":110,"acceptedAnswer":120},"What is PatentNet and what does the paper demonstrate?",{"text":121,"@type":113},"PatentNet is a baseline deep learning model trained on DeepPatent. The paper shows improved retrieval performance over other deep learning approaches and classic computer vision descriptors.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},648045,1792921156,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":14,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":52,"language":139,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":67,"update_tm":143,"read_time":144},8796095461564,"https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d","DeepPatent: Large scale patent drawing recognition and retrieval  \nMichal Kucer, Diane Oyen, Juan Castorena Los Alamos National Laboratory Los Alamos, NM  \n{michal, doyen, [jcastorena](jcastorena}@lanl.gov)[}](jcastorena}@lanl.gov)[@lanl.gov](jcastorena}@lanl.gov)  \nJian Wu Old Dominion University  \nNorfolk, VA [jwu@cs.odu.edu](jwu@cs.odu.edu)  \nAbstract  \nWe tackle the problem of analyzing and retrieving technical drawings. First, we introduce DeepPatent, a new largescale dataset for recognition and retrieval of design patent drawings. The dataset provides more than 350,000 design patent drawings for the purpose of image retrieval. Unlike existing datasets, DeepPatent provides fine-grained image retrieval associations within the collection of drawings and does not rely on cross-domain associations for supervision. We develop a baseline deep learning model, named PatentNet, based on best practices for training retrieval models for static images. We demonstrate the superior performance of PatentNet when trained on our fine-grained associations of DeepPatent against other deep learning approaches and classic computer vision descriptors. With the introduction of this new dataset, and benchmark algorithms, we demonstrate that the analysis and retrieval of technical drawings remains an open challenge in computer vision; and that patent drawing retrieval provides a real-world testbench to spur research.  \n1. Introduction  \nDrawings, illustrations, and free-hand-sketches are often used to convey important scientific or technical information more easily than can be described in text [19, 39] . Research indicates that humans can learn faster and gain deeper understanding from carefully constructed illustration, as opposed to text alone [5, 29] . A technical drawing 1 is a visual description of an object or concept, conveying important information to a person who does not need to have specific expertise to understand the image [14] . Technical and scientific illustration remain a vital part of conveying information in science and technology [19]; especially in archaeology [30,42], medicine [16,21], design [11], and fashion [20] . Yet, information retrieval for scientific, technical,  \n1We use “drawing” as the most appropriate term in computer vision, whereas “illustration” would be more appropriate in the art community.  \nFigure 1: Mosaic of example drawings from five patents: toy figure, moose-shaped animal toy, spectacles, toy wheel, hairbrush. Each patent contains multiple drawings of a single object from various views. One need not be an expert to match which of these drawings belong to each of the five patents (answer given in Supplement) .  \nand scholarly information relies primarily on text-based retrieval—ignoring the drawings even though the drawings may convey more human-accessible information than the text [4, 28] . Our broad goal is to advance computer vision in this area of understanding visual information which is specifically created for human cognition of abstract, technical, and scientific concepts.  \nTo better understand technical drawings, we focus on patent drawings, as in Figure 1, which are a rich yet not very well explored domain [36, 46, 52] . Patent drawings  \nare similar to free-hand sketches in that they are both drawings and thus share some properties: lack of background or any contextual information present in natural photos, abstractness, sparseness, etc [49] . In contrast, patents are often more detailed and of higher quality than free-hand sketches (existing sketch datasets self-describe as “badly drawn bunnies” [39]); giving a faithful—rather than exaggerated—representation of an object [14] . Furthermore, patents typically provide a drawing of an object from several different viewpoints including viewpoints that would rarely be featured in a free-hand-sketch (such as undersides and aerial views) [39]; much as a photo could be taken of an object from any viewpoint. We find that commercial image retrieval","cbCaiqQr496boZce","https://ap.wps.com/l/cbCaiqQr496boZce","pdf",2224967,"English","# Introduction\n## Technical drawings and why text-only retrieval is insufficient\n## Patent drawings as a challenging domain\n## DeepPatent dataset and fine-grained within-domain associations\n## PatentNet baseline and benchmarking","[{\"question\":\"What problem does DeepPatent target?\",\"answer\":\"DeepPatent focuses on analyzing and retrieving technical drawings, particularly design patent drawings, using visual content rather than text-only methods.\"},{\"question\":\"What distinguishes the DeepPatent dataset from existing datasets?\",\"answer\":\"DeepPatent provides fine-grained image retrieval associations within the patent drawing collection and does not rely on cross-domain supervision.\"},{\"question\":\"What is PatentNet and what does the paper demonstrate?\",\"answer\":\"PatentNet is a baseline deep learning model trained on DeepPatent. The paper shows improved retrieval performance over other deep learning approaches and classic computer vision descriptors.\"}]","DeepPatent - Large-scale Patent Drawing Recognition and Retrieval - Paper | PDF",1792791864,25]