[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-650321-105":59,"doc-detail-650321-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","segmenting-technical-drawing-figures-in-us-patents","Segmenting Technical Drawing Figures in US Patents","","Segmentation is central to computer vision for recognizing objects at the pixel level, but dense labeling needs strong contextual modeling. Prior work largely targets natural images, leaving sketched drawings and technical patent figures underexplored. This study presents heuristic point-shooting and deep learning approaches (U-Net, HR-Net, MedT, DETR) for US patent technical drawings. Experiments on a US Patent dataset achieve over 90% accuracy, with transformer-based MedT reaching about 97% segmentation accuracy.",{"@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/segmenting-technical-drawing-figures-in-us-patents/650321/",{"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/segmenting-technical-drawing-figures-in-us-patents/650321.png","ImageObject",300,407,{"name":92,"@type":93},"Putri","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-30","2026-10-24",true,{"@type":102,"interactionType":103,"userInteractionCount":24},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"Why is segmenting technical drawings in US patents difficult?","Question",{"text":112,"@type":113},"Pixel-level segmentation needs contextual information, and patent drawings contain compound figures with subfigures, labels, and references that complicate associating components correctly.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"What segmentation methods are proposed in the study?",{"text":117,"@type":113},"The work introduces a heuristic point-shooting method and compares it with deep learning models including U-Net, HR-Net, MedT, and DETR.",{"name":119,"@type":110,"acceptedAnswer":120},"How accurate are the proposed models on the US Patent dataset?",{"text":121,"@type":113},"The methods achieve over 90% accuracy on the dataset; transformer-based MedT performs best, reaching about 97% segmentation accuracy.","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},650321,1792921513,{"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":24,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":29,"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},962085571259,"https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0","Segmenting Technical Drawing Figures in US Patents  \nMd Reshad Ul Hoque, 1 Xin Wei,2 Muntabir Hasan Choudhury,2 Kehinde Ajayi,2 Martin Gryder,2  \nJian Wu,2 Diane Oyen3  \n1Electrical and Computer Engineering, Old Dominion University  \n[f](fmhoqu001g@odu.edu)[mhoqu001](fmhoqu001g@odu.edu)[g](fmhoqu001g@odu.edu)[@odu.edu](fmhoqu001g@odu.edu)  \n2 Computer Science, Old Dominion University  \nfxwei001,mchou001,kajay001,[j1wu](j1wug@odu.edu)[g](j1wug@odu.edu)[@odu.edu](j1wug@odu.edu)  \n2Los Alamos National Laboratory  \n[doyen@lanl.gov](doyen@lanl.gov)  \nAbstract  \nImage segmentation is the core computer vision problem for identifying objects within a scene. Segmentation is a challenging task because the prediction for each pixel label requires contextual information. Most recent research deals with the segmentation of natural images rather than drawings. However, there is very little research on sketched image segmentation. In this study, we introduce heuristic (pointshooting) and deep learning-based methods (U-Net, HR-Net, MedT, DETR) to segment technical drawings in US patent documents. Our proposed methods on the US Patent dataset achieved over 90% accuracy where transformer performs well with 97% segmentation accuracy, which is promising and computationally efficient. Our source codes and datasets are available at [https://github.com/reshadshuvo123/](https://github.com/reshadshuvo123/)[ ](https://github.com/reshadshuvo123/)[Sketched-Image-Segmentation](Sketched-Image-Segmentation)  \nIntroduction  \nCombining information contained in text and images isan important aspect of understanding scientific documents. However, patents and scientific documents often contain compound figures containing subfigures, each having its own label, caption, and reference text. To associate individual subfigures with the appropriate caption and reference text, we must first segment the full figure into its individual subfigures. Although much research has been done on figure understanding and extraction for scientific documents, existing methods rely on either (1) manually-designed rules and human-crafted features which do not generalize well for new dataset (Taschwer and Marques 2018); or (2) machine learning approaches most of which were trained on natural images (Tsutsui and Crandall 2017) . We demonstrate that we cannot simply apply approaches developed for other datasets to patent drawings, and develop a novel approach while addressing questions about how to extend existing methods to novel datasets.  \nImage segmentation has been extensively studied with rule-based methods such as watershed, and machine learning methods applied to natural images (Bai and Urtasun 2017; Ren and Zemel 2017; Minaee et al. 2021) . In patent drawings, there are usually white space between individual drawings. A simple sweeping line method, which detect  \nCopyright © 2022, Association for the Advancement of Artificial Intelligence ([www.aaai.org](www.aaai.org)). All rights reserved.  \nboundaries of subfigures by counting the maximum number of black-pixels along a horizontal (or vertical) pixel array, e.g., Rane et al. (2021), does not work well because (1) other components such as figure labels may be present, and (2) one figure may contain multiple disconnected parts. There are few existing papers on segmenting technical drawings in patent documents. Most existing tools were developed for extracting figures in research papers. For example, Clark and Divvala (2016) developed a framework that extracts figures from scientific papers in PDF format. Viziometrics (Lee, West, and Howe 2017), a figure-oriented literature mining system, was developed which works on certain pattern figures. These tools could not be used for segmenting compound figures. For compound figure separation, background color, layout patterns, spaces and lines between sub figures were used as important cues for rule-based methods (Taschwer and Marques 2018). Tsutsui [el. at. developed a](el. at. developed a)[ ](el. at. develope","cbCaisikw3k70tCq","https://ap.wps.com/l/cbCaisikw3k70tCq","pdf",7557309,"English","# Introduction\n# Abstract\n# Data\n# Methods and Results\n## Heuristic point-shooting\n## Deep learning models\n## Transformer-based segmentation","[{\"question\":\"Why is segmenting technical drawings in US patents difficult?\",\"answer\":\"Pixel-level segmentation needs contextual information, and patent drawings contain compound figures with subfigures, labels, and references that complicate associating components correctly.\"},{\"question\":\"What segmentation methods are proposed in the study?\",\"answer\":\"The work introduces a heuristic point-shooting method and compares it with deep learning models including U-Net, HR-Net, MedT, and DETR.\"},{\"question\":\"How accurate are the proposed models on the US Patent dataset?\",\"answer\":\"The methods achieve over 90% accuracy on the dataset; transformer-based MedT performs best, reaching about 97% segmentation accuracy.\"}]","Segmenting Technical Drawing Figures in US Patents | PDF",1792824888,15]