[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-86063-en":3,"doc-seo-86063-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},86063,687197207057,"Sage","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","Align and Segment: Unsupervised Learning for Building Segmentation From Misaligned Labels","Spatially aligned image–label pairs are typically required for supervised image segmentation, but remote sensing data often contains systematic label noise due to misalignment between independently sourced imagery and masks. AnS (Align and Segment) jointly learns an affine alignment of misaligned labels and canonical semantic segmentation targets using a spatial transformer module. Self-supervised regularization prevents shortcut learning, complementing data augmentation. The method trains without any “golden” clean labels and generalizes across cities and varying noise levels for building footprint segmentation.","Align and Segment: Unsupervised Learning for Building Segmentation From Misaligned Labels  \n[ cs .CV] 12 Jul 2026  \nVenkanna Babu Guthula 1, Oswin Krause 1, Dimitri Gominski 1, Hui Zhang 1, Johan Mottelson2, Ankit Kariryaa 1, Nico Lang 1, and  \nChristian Igel 1  \n1 University of Copenhagen, Copenhagen, Denmark  \n2 Royal Danish Academy, Copenhagen, Denmark  \n{vegu, [igel}@di.ku.dk](igel}@di.ku.dk)  \nAbstract. Supervised learning for image segmentation typically requires spatially aligned image and label sets. When images and labels originate from different sources, the pairing may be misaligned, which can significantly deteriorate the performance of the learned models. This is especially common in remote sensing, when aerial or satellite images are co-registered with labels from another source (e.g., OpenStreetMap) . In this work, we propose a novel approach for training on misaligned labels, where we simultaneously learn the label alignment. Our align and segment (AnS) approach builds on the spatial transformer module to transform the misaligned labels using an affine transformation to provide a better learning target for a canonical semantic segmentation network. We prevent shortcut learning of misaligned labels in these semantic segmentation networks through a self-supervised regularization loss and show that it is complementary to data augmentation, especially for systematically misaligned training data. A decisive characteristic of our AnS approach is that it learns without requiring any “golden” labels. We exper-  \n2 Guthula et al.  \nFig. 1: Misalignment of building footprints with independent imagery. Data e.g. from Google Open Buildings [27] or OpenStreetMap (OSM) [23] neither align with independent satellite imagery nor with each other. The right figure shows our estimated misalignment of OSM data, highlighting the locally systematic label noise. The arrows indicate estimated translations, with color representing the displacement magnitude.  \nIn remote sensing, where images (e.g., satellite or drone imagery) and labels (e.g., crowd-sourced or field-collected) often originate from different sources, label noise is particularly common, especially spatial misalignment of images and label masks. For example, overlaying OpenStreetMap (OSM) [23] building footprints against other public building maps reveals large and systematic local transformations of several pixels (Fig. 1) . This bottleneck has led to an underutilization of large swaths of hand-curated data such as OSM for training deep learning models at scale and in regions where high-quality building data are scarce.  \nAt first glance, jointly modeling label alignment and segmentation may appear to be a straightforward solution. However, such a disentanglement is nontrivial and a naive implementation in two sub-networks is prone to fail, due to the ability of segmentation networks to simply learn the misaligned labels. An alternative popular approach is to view the problem as a multimodal image alignment task and to estimate the transformation between the image and the label maps using feature maps learned for the different modalities [8, 9, 31] . However, these approaches usually assume that the observed misalignments in the dataset are small and unbiased. In remote sensing these assumptions often do not hold, as can be seen in Fig. 1, where large and biased misalignments of images can be introduced by erroneous orthorectification (i.e., projecting images onto terrain) and other satellite image preprocessing steps as well as errors made in the land registry offices.  \nIn this work, we propose a learning-based approach for dealing with misaligned labels in segmentation without requiring any “golden” (clean, groundtruth) labels. Our solution involves several components of regularization, in-  \nAlign and Segment 3  \ncluding a new loss that allows two sub-networks to learn label alignment and segmentation respectively. We focus on the segmentation of building footprints f","cbCaijDoG8RClFzs","https://ap.wps.com/l/cbCaijDoG8RClFzs","pdf",26357683,2,1,29,"English","en",105,"# Abstract\n# Introduction\n## Misalignment and label noise in remote sensing\n## Challenge of joint alignment and segmentation\n# Method: Align and Segment (AnS)\n## Self-consistency loss and regularization\n## Learning without golden labels\n# Contributions\n# Related work","[{\"question\":\"Why do misaligned labels harm supervised image segmentation performance?\",\"answer\":\"When images and labels originate from different sources, their pairing can be spatially misaligned. This label noise deteriorates the learned model’s performance because the network learns incorrect correspondences between pixels and targets.\"},{\"question\":\"How does the Align and Segment (AnS) approach learn from misaligned labels without golden annotations?\",\"answer\":\"AnS uses a spatial transformer to learn an affine transformation that aligns misaligned labels to a canonical semantic segmentation target. A self-supervised regularization loss discourages the segmentation network from simply learning the misaligned labels, enabling training without clean ground-truth labels.\"},{\"question\":\"What problem settings does the paper focus on, and what application motivates the method?\",\"answer\":\"The method targets building footprint segmentation from high-resolution satellite imagery. It is motivated by urban planning needs and disease/disaster risk assessment in informal settlements, where curated high-quality labeled data is scarce and open maps like OSM require extensive alignment and correction.\"}]",1784208209,73,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"align-and-segment-unsupervised-learning-for-building-segmentation-from-misaligned-labels","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":20},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/align-and-segment-unsupervised-learning-for-building-segmentation-from-misaligned-labels/86063/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-18","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why do misaligned labels harm supervised image segmentation performance?","Question",{"text":75,"@type":76},"When images and labels originate from different sources, their pairing can be spatially misaligned. This label noise deteriorates the learned model’s performance because the network learns incorrect correspondences between pixels and targets.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the Align and Segment (AnS) approach learn from misaligned labels without golden annotations?",{"text":80,"@type":76},"AnS uses a spatial transformer to learn an affine transformation that aligns misaligned labels to a canonical semantic segmentation target. A self-supervised regularization loss discourages the segmentation network from simply learning the misaligned labels, enabling training without clean ground-truth labels.",{"name":82,"@type":73,"acceptedAnswer":83},"What problem settings does the paper focus on, and what application motivates the method?",{"text":84,"@type":76},"The method targets building footprint segmentation from high-resolution satellite imagery. It is motivated by urban planning needs and disease/disaster risk assessment in informal settlements, where curated high-quality labeled data is scarce and open maps like OSM require extensive alignment and correction.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]