{"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/proposal-flow-semantic-correspondences-from","title":"Proposal Flow: Semantic Correspondences from Object Proposals","arxiv_id":"1703.07144","date":"2017-03-21","proceeding":null,"authors":["Bumsub Ham","Minsu Cho","Cordelia Schmid","Jean Ponce"],"abstract":"Finding image correspondences remains a challenging problem in the presence\nof intra-class variations and large changes in scene layout. Semantic flow\nmethods are designed to handle images depicting different instances of the same\nobject or scene category. We introduce a novel approach to semantic flow,\ndubbed proposal flow, that establishes reliable correspondences using object\nproposals. Unlike prevailing semantic flow approaches that operate on pixels or\nregularly sampled local regions, proposal flow benefits from the\ncharacteristics of modern object proposals, that exhibit high repeatability at\nmultiple scales, and can take advantage of both local and geometric consistency\nconstraints among proposals. We also show that the corresponding sparse\nproposal flow can effectively be transformed into a conventional dense flow\nfield. We introduce two new challenging datasets that can be used to evaluate\nboth general semantic flow techniques and region-based approaches such as\nproposal flow. We use these benchmarks to compare different matching\nalgorithms, object proposals, and region features within proposal flow, to the\nstate of the art in semantic flow. This comparison, along with experiments on\nstandard datasets, demonstrates that proposal flow significantly outperforms\nexisting semantic flow methods in various settings.","url_abs":"http://arxiv.org/abs/1703.07144v1","url_pdf":"http://arxiv.org/pdf/1703.07144v1.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":[],"tasks":[{"task_slug":"object","task_name":"Object"}],"methods":[],"datasets_introduced":[{"slug":"pf-pascal","name":"PF-PASCAL","full_name":""},{"slug":"pf-willow","name":"PF-WILLOW","full_name":""}],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1703.07144","atlas_url":"https://app.syntology.ai/?focus=1703.07144","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}