[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-121900-en":3,"doc-seo-121900-105":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":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},121900,4398048949847,"Eliana","https://ap-avatar.wpscdn.com/avatar/400002536579ef2da7f?_k=1778318612642679267",8,"Research & Report","RASP - Relation-aware Semantic Prior for Weakly Supervised Incremental Segmentation","Class-incremental semantic image segmentation updates a model to learn new categories while retaining old ones. Dense pixel-level annotations for every newly introduced object are expensive and limit real-world adoption. Image-level labels reduce annotation effort but provide weak localization and unclear object boundaries. This work proposes Relation-aware Semantic Prior (RaSP), transferring objectness cues from previously learned classes to new ones by exploiting semantic relations between class labels, improving mask quality for both old and new categories across continual and few-shot scenarios.","RASP: RELATION-AWARE SEMANTIC PRIOR FOR WEAKLY SUPERVISED INCREMENTAL SEGMENTATION  \nSubhankar Roy  \nLTCI, Tlcom-Paris  \nIntitute Polytechnique de Paris [subhankar.roy@telecom-paris.fr](subhankar.roy@telecom-paris.fr)  \nRiccardo Volpi, Gabrela Csurka, Diane Larlus  \nNAVER LABS Europe  \nMeylan, France  \n{[name.lastname](name.lastname}@naverlabs.com)[}](name.lastname}@naverlabs.com)[@naverlabs.com](name.lastname}@naverlabs.com)  \nABSTRACT  \nClass-incremental semantic image segmentation assumes multiple model updates, each enriching the model to segment new categories. This is typically carried out by providing expensive pixellevel annotations to the training algorithm for all new objects, limiting the adoption of such methods in practical applications. Approaches that solely require image-level labels offer an attractive alternative, yet, such coarse annotations lack precise information about the location and boundary of the new objects. In this paper we argue that, since classes represent not just indices but semantic entities, the conceptual relationships between them can provide valuable information that should be leveraged. We propose a weakly supervised approach that exploits such semantic relations to transfer objectness prior from the previously learned classes into the new ones, complementing the supervisory signal from image-level labels. We validate our approach on a number of continual learning tasks, and show how even a simple pairwise interaction between classes can significantly improve the segmentation mask quality of both old and new classes. We show these conclusions still hold for longer and, hence, more realistic sequences of tasks and for a challenging few-shot scenario.  \n1 INTRODUCTION  \nWhen working towards the real-world deployment of artificial intelligence systems, two main challenges arise: such systems should possess the ability to continuously learn, and this learning process should only require limited human intervention. While deep learning models have proved effective in tackling tasks for which large amounts of curated data as well as abundant computational resources are available, they still struggle to learn over continuous and potentially heterogeneous sequences of tasks, especially if supervision is limited.  \nIn this work, we focus on the task of semantic image segmentation (SIS) (Csurka et al., 2022), where the goal is predicting the class label of each pixel in an image. A reliable and versatile SIS model should be able to seamlessly add new categories to its repertoire without forgetting about the old ones. Considering for instance a house robot or a self-driving vehicle with such segmentation capability, we would like it to extend its knowledge to new classes without having to retrain the segmentation model from scratch on the old ones. Such ability is at the core of continual learning research, the main challenge being to mitigate catastrophic forgetting of what has been previously learned (Parisi et al., 2019) .  \nMost learning algorithms for SIS assume training samples with dense pixel-level annotations, an ex-  \nFigure 1: Our proposed Relation-aware Semantic Prior (RaSP) loss is based on the intuition that old class predictions from an existing model provide valuable cues for the segmentation of unseen, semantically related classes. Based on the semantic relatedness of the image label (e.g., sheep) and the model predictions (e.g., cow), our model derives denser maps for the new class and leverages them during training  \npensive and tedious operation. We argue that this is cumbersome and severely hinders continual learning; adding new classes over time should be an annotation friendly process. This is why, here, we focus on the case where only image-level labels are provided (e.g., adding the ‘sheep’ class comes as easily as only providing images guaranteed to contain at least a sheep) . However, this task, denoted as Weakly Supervised Class-Incremental (WSCI) SIS, is an  \nextremely c","cbCaidtw0cGqYlXW","https://ap.wps.com/l/cbCaidtw0cGqYlXW","pdf",26263637,1,26,"English","en",105,"# Abstract\n# Introduction\n## Continual learning and semantic image segmentation\n## Weakly supervised class-incremental setting\n## Relation-aware Semantic Prior (RaSP) idea\n## RaSP loss motivation and generality","[{\"question\":\"What problem does class-incremental semantic image segmentation address?\",\"answer\":\"It enables adding new semantic categories to a segmentation model over time while mitigating forgetting of previously learned categories.\"},{\"question\":\"Why do image-level labels make incremental segmentation difficult?\",\"answer\":\"Image-level supervision lacks precise information about where new objects are located and what their boundaries are, leading to coarse and inaccurate masks.\"},{\"question\":\"How does RaSP improve weakly supervised incremental segmentation?\",\"answer\":\"RaSP uses semantic relationships between class names to transfer objectness priors from old classes, converting coarse image-level signals into dense pixel-level training cues.\"}]","RASP - Relation-aware Semantic Prior for Weakly Supervised Incremental Segmentation | PDF",1785807646,66,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"rasp-relation-aware-semantic-prior-for-weakly-supervised-incremental-segmentation","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/rasp-relation-aware-semantic-prior-for-weakly-supervised-incremental-segmentation/121900/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-04",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does class-incremental semantic image segmentation address?","Question",{"text":75,"@type":76},"It enables adding new semantic categories to a segmentation model over time while mitigating forgetting of previously learned categories.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Why do image-level labels make incremental segmentation difficult?",{"text":80,"@type":76},"Image-level supervision lacks precise information about where new objects are located and what their boundaries are, leading to coarse and inaccurate masks.",{"name":82,"@type":73,"acceptedAnswer":83},"How does RaSP improve weakly supervised incremental segmentation?",{"text":84,"@type":76},"RaSP uses semantic relationships between class names to transfer objectness priors from old classes, converting coarse image-level signals into dense pixel-level training cues.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"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"]