{"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/spatial-mixture-models-with-learnable-deep","title":"Spatial Mixture Models with Learnable Deep Priors for Perceptual Grouping","arxiv_id":"1902.02502","date":"2019-02-07","proceeding":null,"authors":["Jinyang Yuan","Bin Li","xiangyang xue"],"abstract":"Humans perceive the seemingly chaotic world in a structured and compositional\nway with the prerequisite of being able to segregate conceptual entities from\nthe complex visual scenes. The mechanism of grouping basic visual elements of\nscenes into conceptual entities is termed as perceptual grouping. In this work,\nwe propose a new type of spatial mixture models with learnable priors for\nperceptual grouping. Different from existing methods, the proposed method\ndisentangles the attributes of an object into ``shape'' and ``appearance''\nwhich are modeled separately by the mixture weights and the mixture components.\nMore specifically, each object in the visual scene is fully characterized by\none latent representation, which is in turn transformed into parameters of the\nmixture weight and the mixture component by two neural networks. The mixture\nweights focus on modeling spatial dependencies (i.e., shape) and the mixture\ncomponents deal with intra-object variations (i.e., appearance). In addition,\nthe background is separately modeled as a special component complementary to\nthe foreground objects. Our extensive empirical tests on two perceptual\ngrouping datasets demonstrate that the proposed method outperforms the\nstate-of-the-art methods under most experimental configurations. The learned\nconceptual entities are generalizable to novel visual scenes and insensitive to\nthe diversity of objects. Code is available at\nhttps://github.com/jinyangyuan/learnable-deep-priors.","url_abs":"http://arxiv.org/abs/1902.02502v2","url_pdf":"http://arxiv.org/pdf/1902.02502v2.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":"spatial-mixture-models-with-learnable-deep","repo_url":"https://github.com/jinyangyuan/learnable-deep-priors","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"object","task_name":"Object"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1902.02502","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}