{"id":37437,"date":"2025-12-25T17:27:10","date_gmt":"2025-12-25T16:27:10","guid":{"rendered":"https:\/\/www.graviton.at\/letterswaplibrary\/dataset-release-real-structural-engineering-drawings-for-ai-pned-6-rc-datasets\/"},"modified":"2025-12-25T17:27:10","modified_gmt":"2025-12-25T16:27:10","slug":"dataset-release-real-structural-engineering-drawings-for-ai-pned-6-rc-datasets","status":"publish","type":"post","link":"https:\/\/www.graviton.at\/letterswaplibrary\/dataset-release-real-structural-engineering-drawings-for-ai-pned-6-rc-datasets\/","title":{"rendered":"Dataset Release: Real Structural Engineering Drawings For AI (PNED \u2013 6 RC Datasets)"},"content":{"rendered":"<p><!-- SC_OFF --><\/p>\n<div class=\"md\">\n<p>Hi everyone,<\/p>\n<p>I\u2019ve been working as a structural engineer for about 10 years (Germany, RC design).<br \/> Over the last few years I\u2019ve noticed something very surprising in AI\/ML:<\/p>\n<p>We have datasets for almost everything \u2014 but none for real structural engineering drawings.<\/p>\n<p>These drawings are extremely challenging for machine learning due to:<\/p>\n<ul>\n<li>dense, overlapping geometry<\/li>\n<li>structural symbols and reinforcement notation<\/li>\n<li>dimensions, leaders, section markers<\/li>\n<li>multi-layer technical detailing<\/li>\n<li>scale-dependent information<\/li>\n<li>mixed text + geometry + symbols<\/li>\n<\/ul>\n<p>Because of this, they are highly relevant for:<\/p>\n<ul>\n<li>OCR \/ document understanding<\/li>\n<li>object detection<\/li>\n<li>layout analysis<\/li>\n<li>symbol recognition<\/li>\n<li>segmentation<\/li>\n<li>BIM automation<\/li>\n<li>engineering-focused CV research<\/li>\n<\/ul>\n<p>So I started building a series of datasets of real reinforced-concrete drawings, created specifically for ML tasks.<\/p>\n<p>Each dataset contains:<\/p>\n<ul>\n<li>25 PDF engineering drawings (Columns 50 PDF)<\/li>\n<li>25 PNG images (1200 dpi) (Columns 50 PDF)<\/li>\n<li>one structural category per dataset (RC beams, walls, foundations, columns, precast columns, etc.)<\/li>\n<\/ul>\n<p>So far I\u2019ve released 6 datasets:<\/p>\n<ul>\n<li>RC Beams V1<\/li>\n<li>RC Columns V1<\/li>\n<li>RC Foundations V1<\/li>\n<li>RC Precast Columns V1<\/li>\n<li>RC Walls V1<\/li>\n<li>RC Walls V2<\/li>\n<\/ul>\n<p>All datasets, including sample images, can be viewed here:<\/p>\n<p>\ud83d\udc49 [<a href=\"https:\/\/huggingface.co\/PNEngineeringDatasets%5D()\">https:\/\/huggingface.co\/PNEngineeringDatasets]()<\/a><\/p>\n<p>I\u2019d be happy to hear any feedback, suggestions or use cases you think could be valuable for ML research in this domain.<\/p>\n<p>Disclaimer: this is my own dataset project; posting once for visibility.<\/p>\n<\/div>\n<p><!-- SC_ON -->   submitted by   <a href=\"https:\/\/www.reddit.com\/user\/PNEngineeringDataset\"> \/u\/PNEngineeringDataset <\/a> <br \/> <span><a href=\"https:\/\/www.reddit.com\/r\/datasets\/comments\/1pckcvt\/dataset_release_real_structural_engineering\/\">[link]<\/a><\/span>   <span><a href=\"https:\/\/www.reddit.com\/r\/datasets\/comments\/1pckcvt\/dataset_release_real_structural_engineering\/\">[comments]<\/a><\/span><\/p><div class='watch-action'><div class='watch-position align-right'><div class='action-like'><a class='lbg-style1 like-37437 jlk' href='javascript:void(0)' data-task='like' data-post_id='37437' data-nonce='65e0e39b87' rel='nofollow'><img class='wti-pixel' src='https:\/\/www.graviton.at\/letterswaplibrary\/wp-content\/plugins\/wti-like-post\/images\/pixel.gif' title='Like' \/><span class='lc-37437 lc'>0<\/span><\/a><\/div><\/div> <div class='status-37437 status align-right'><\/div><\/div><div class='wti-clear'><\/div>","protected":false},"excerpt":{"rendered":"<p>Hi everyone, I\u2019ve been working as a structural engineer for about 10 years (Germany, RC design). Over&#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[85],"tags":[],"class_list":["post-37437","post","type-post","status-publish","format-standard","hentry","category-datatards","wpcat-85-id"],"_links":{"self":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/posts\/37437","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/comments?post=37437"}],"version-history":[{"count":0,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/posts\/37437\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/media?parent=37437"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/categories?post=37437"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/tags?post=37437"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}