{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/seed-expand-and-constrain-three-principles","title":"Seed, Expand and Constrain: Three Principles for Weakly-Supervised Image Segmentation","arxiv_id":"1603.06098","date":"2016-03-19","proceeding":null,"authors":["Alexander Kolesnikov","Christoph H. Lampert"],"abstract":"We introduce a new loss function for the weakly-supervised training of\nsemantic image segmentation models based on three guiding principles: to seed\nwith weak localization cues, to expand objects based on the information about\nwhich classes can occur in an image, and to constrain the segmentations to\ncoincide with object boundaries. We show experimentally that training a deep\nconvolutional neural network using the proposed loss function leads to\nsubstantially better segmentations than previous state-of-the-art methods on\nthe challenging PASCAL VOC 2012 dataset. We furthermore give insight into the\nworking mechanism of our method by a detailed experimental study that\nillustrates how the segmentation quality is affected by each term of the\nproposed loss function as well as their combinations.","url_abs":"http://arxiv.org/abs/1603.06098v3","url_pdf":"http://arxiv.org/pdf/1603.06098v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"seed-expand-and-constrain-three-principles","repo_url":"https://github.com/kolesman/SEC","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"seed-expand-and-constrain-three-principles","repo_url":"https://github.com/halbielee/SEC_pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"image-segmentation","task_name":"Image Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1603.06098","atlas_url":"https://app.syntology.ai/?focus=1603.06098","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}