[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-657701-105":59,"doc-detail-657701-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","topology-aware-structural-parsing-of-hand-drawn-diagrams-via-learning-aligned-decoding","Topology-Aware Structural Parsing of Hand-Drawn Diagrams via Learning-Aligned Decoding","","Automatic interpretation of hand-drawn technical diagrams is difficult because the target is a directed graph rather than merely visible regions. Visual evidence such as connector appearance can match at pixel level while producing an incorrect source–target relation. A multi-head graph-evidence network predicts nodes, arrow components, direction cues, path progress, and endpoint geometry, then a deterministic two-pass assembler forms a provisional graph and finalizes attachments using graph context. Experiments on 450 held-out diagrams report high node and link F1 scores, strong centerline Dice, and low graph edit distance.",{"@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":30,"@type":76,"position":81},"https://docshare.wps.com/document/technology/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/topology-aware-structural-parsing-of-hand-drawn-diagrams-via-learning-aligned-decoding/657701/",{"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/topology-aware-structural-parsing-of-hand-drawn-diagrams-via-learning-aligned-decoding/657701.png","ImageObject",300,407,{"name":92,"@type":93},"Mia  ","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-30","2026-10-26",true,{"@type":102,"interactionType":103,"userInteractionCount":19},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"Why is parsing hand-drawn technical diagrams harder than ordinary segmentation?","Question",{"text":112,"@type":113},"Because the diagram represents a directed graph, not just visible regions. Small visual mistakes can still cause discrete structural errors in source, target, and direction of edges.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"What does the proposed model predict to enable structural recovery?",{"text":117,"@type":113},"It predicts node regions, arrow shafts and endpoints, local direction, path progress, endpoint offsets, and endpoint tangents using a multi-head graph-evidence network.",{"name":119,"@type":110,"acceptedAnswer":120},"How does the two-pass assembler improve graph extraction reliability?",{"text":121,"@type":113},"It first constructs a provisional physical graph, then uses graph context to complete selected attachments while suppressing unsupported fallback traces and physical duplicates, refining matched node regions.","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},657701,1793051323,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":29,"category_name":30,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":19,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"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":145},687207024478,"https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd","Topology-Aware Structural Parsing of Hand-Drawn Diagrams via Learning-Aligned Decoding  \nHrishikesh Vichore  \nVisvesvaraya National Institute of Technology  \nMansi Radke  \nVisvesvaraya National Institute of Technology  \nPraveen Kumar  \nVisvesvaraya National Institute of Technology  \nResearch Article  \nKeywords: handwritten diagram parsing, visual-to-graph recovery, topology-aware decoding, directed graph extraction, structural assembly, document image analysis  \nPosted Date: August 13th, 2026  \nDOI: [https://doi.org/10.21203/rs.3.rs-10534078/v1](https://doi.org/10.21203/rs.3.rs-10534078/v1)  \nLicense: 􀁱 􀅏 This work is licensed under a Creative Commons Attribution 4.0 International License.  \nRead Full License  \nAdditional Declarations: No competing interests reported.  \nTopology-Aware Structural Parsing of Hand-Drawn Diagrams via Learning-Aligned Decoding  \nHrishikesh Vichore 1*, Mansi Radke 1 and Praveen Kumar 1  \n1 Department of Computer Science and Engineering, Visvesvaraya National Institute of Technology, Nagpur, Maharashtra, India.  \n*Corresponding author(s). E-mail(s): [dt24cse024@students.vnit.ac.in](dt24cse024@students.vnit.ac.in) ;  \nAbstract  \nPurpose: Automatic interpretation of hand-drawn technical diagrams is difficult because the target is a directed graph rather than a collection of visible regions. A connector may appear nearly correct atthe pixel level and still produce a wrong source–target relation. We study this gap between visual evidence and structural recovery. Methods: A multi-head graph-evidence network predicts node regions, arrow shafts, skeletons, endpoints, local direction, path progress, endpoint offsets, and endpoint tangents. Long-arrow-aware supervision reduces fragmentation in extended connectors, while a separate 34-layer residual network refines symbol regions. A deterministic two-pass assembler first constructsa provisional physical graph and then uses graph context to complete selected attachments, suppress unsupported fallback traces and physical duplicates, recover strongly supported structural deficits, and refine matched node regions. Results: On 450 held-out diagrams containing finite automata and flowcharts, the complete framework obtains a node F1 score of 98.57%, mask Intersection over Union (IoU) of 82.71%, soft centerline Dice (Soft-clDice) of 96.72%, a 3.77-pixel endpoint error, a directed Link F1 score of 92.49%, and a normalized Graph Edit Distance (nGED) of 0.090 . Loop-back and decision-branch recall reach 95 .24% and 94 .04%, respectively. Conclusion: Learning-aligned fields make connector evidence recoverable, but reliable graph extraction also requires delaying ambiguous structural decisions until a provisional graph exists. The two-pass formulation provides that context without introducing a separate semantic reasoner.  \nKeywords: handwritten diagram parsing, visual-to-graph recovery, topology-aware decoding, directed graph extraction, structural assembly, document image analysis  \n1 Introduction  \nHand-drawn technical diagrams are compact descriptions of structure. In engineering education, programming assessment, and document image analysis, a flowchart or a finite automaton is not merely an illustration placed beside text. It encodes a graph. Vertices correspond to states, symbols, decisions, or process blocks, and edges define directed relations between them.  \nA useful parser must therefore recover not only where visual components lie, but also how those components are connected.  \nThis makes the problem less forgiving than ordinary segmentation. A shaft mask can overlap the ground truth reasonably well and still be unusable if it breaks at one critical point. A node can be localized correctly, yet the graph may be wrong if an arrow endpoint snaps to the wrong side of it. This seems counterintuitive at first because the  \nvisual error may be small. From the graph point of view, however, the error is discrete: an edge is either recovered with the correct source, target, and direc","cbCaijsrYqyVjB6W","https://ap.wps.com/l/cbCaijsrYqyVjB6W","pdf",785868,21,"English","# Abstract\n# 1 Introduction\n## Graph recovery vs segmentation\n## Detect-and-decode limitations\n## Learning-aligned prediction for decoding","[{\"question\":\"Why is parsing hand-drawn technical diagrams harder than ordinary segmentation?\",\"answer\":\"Because the diagram represents a directed graph, not just visible regions. Small visual mistakes can still cause discrete structural errors in source, target, and direction of edges.\"},{\"question\":\"What does the proposed model predict to enable structural recovery?\",\"answer\":\"It predicts node regions, arrow shafts and endpoints, local direction, path progress, endpoint offsets, and endpoint tangents using a multi-head graph-evidence network.\"},{\"question\":\"How does the two-pass assembler improve graph extraction reliability?\",\"answer\":\"It first constructs a provisional physical graph, then uses graph context to complete selected attachments while suppressing unsupported fallback traces and physical duplicates, refining matched node regions.\"}]","Topology-Aware Structural Parsing of Hand-Drawn Diagrams via Learning-Aligned Decoding | PDF",1792986434,53]