{"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/unsupervised-meta-learning-of-figure-ground","title":"Unsupervised Meta-learning of Figure-Ground Segmentation via Imitating Visual Effects","arxiv_id":"1812.08442","date":"2018-12-20","proceeding":null,"authors":["Ding-Jie Chen","Jui-Ting Chien","Hwann-Tzong Chen","Tyng-Luh Liu"],"abstract":"This paper presents a \"learning to learn\" approach to figure-ground image\nsegmentation. By exploring webly-abundant images of specific visual effects,\nour method can effectively learn the visual-effect internal representations in\nan unsupervised manner and uses this knowledge to differentiate the figure from\nthe ground in an image. Specifically, we formulate the meta-learning process as\na compositional image editing task that learns to imitate a certain visual\neffect and derive the corresponding internal representation. Such a generative\nprocess can help instantiate the underlying figure-ground notion and enables\nthe system to accomplish the intended image segmentation. Whereas existing\ngenerative methods are mostly tailored to image synthesis or style transfer,\nour approach offers a flexible learning mechanism to model a general concept of\nfigure-ground segmentation from unorganized images that have no explicit\npixel-level annotations. We validate our approach via extensive experiments on\nsix datasets to demonstrate that the proposed model can be end-to-end trained\nwithout ground-truth pixel labeling yet outperforms the existing methods of\nunsupervised segmentation tasks.","url_abs":"http://arxiv.org/abs/1812.08442v1","url_pdf":"http://arxiv.org/pdf/1812.08442v1.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":"unsupervised-meta-learning-of-figure-ground","repo_url":"https://github.com/LiangHann/USAR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"unsupervised-meta-learning-of-figure-ground","repo_url":"https://github.com/timy90022/VEGAN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"unsupervised-meta-learning-of-figure-ground","repo_url":"https://github.com/timy90022/WEGAN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"image-generation","task_name":"Image Generation"},{"task_slug":"image-segmentation","task_name":"Image Segmentation"},{"task_slug":"meta-learning","task_name":"Meta-Learning"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"style-transfer","task_name":"Style Transfer"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}