{"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/multi-view-dynamic-facial-action-unit","title":"Multi-View Dynamic Facial Action Unit Detection","arxiv_id":"1704.07863","date":"2017-04-25","proceeding":null,"authors":["Andres Romero","Juan Leon","Pablo Arbelaez"],"abstract":"We propose a novel convolutional neural network approach to address the\nfine-grained recognition problem of multi-view dynamic facial action unit\ndetection. We leverage recent gains in large-scale object recognition by\nformulating the task of predicting the presence or absence of a specific action\nunit in a still image of a human face as holistic classification. We then\nexplore the design space of our approach by considering both shared and\nindependent representations for separate action units, and also different CNN\narchitectures for combining color and motion information. We then move to the\nnovel setup of the FERA 2017 Challenge, in which we propose a multi-view\nextension of our approach that operates by first predicting the viewpoint from\nwhich the video was taken, and then evaluating an ensemble of action unit\ndetectors that were trained for that specific viewpoint. Our approach is\nholistic, efficient, and modular, since new action units can be easily included\nin the overall system. Our approach significantly outperforms the baseline of\nthe FERA 2017 Challenge, with an absolute improvement of 14% on the F1-metric.\nAdditionally, it compares favorably against the winner of the FERA 2017\nchallenge. Code source is available at https://github.com/BCV-Uniandes/AUNets.","url_abs":"http://arxiv.org/abs/1704.07863v2","url_pdf":"http://arxiv.org/pdf/1704.07863v2.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":"multi-view-dynamic-facial-action-unit","repo_url":"https://github.com/BCV-Uniandes/AUNets","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"action-unit-detection","task_name":"Action Unit Detection"},{"task_slug":"facial-action-unit-detection","task_name":"Facial Action Unit Detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/facial-action-unit-detection-on-bp4d","task":"Facial Action Unit Detection","dataset":"BP4D","model":"Multi-View Dynamic Facial Action Unit Detection","rank_in_archive_order":6,"of":10,"metrics":{"Average F1":"63.0"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}