{"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/red-reinforced-encoder-decoder-networks-for","title":"RED: Reinforced Encoder-Decoder Networks for Action Anticipation","arxiv_id":"1707.04818","date":"2017-07-16","proceeding":null,"authors":["Jiyang Gao","Zhenheng Yang","Ram Nevatia"],"abstract":"Action anticipation aims to detect an action before it happens. Many real\nworld applications in robotics and surveillance are related to this predictive\ncapability. Current methods address this problem by first anticipating visual\nrepresentations of future frames and then categorizing the anticipated\nrepresentations to actions. However, anticipation is based on a single past\nframe's representation, which ignores the history trend. Besides, it can only\nanticipate a fixed future time. We propose a Reinforced Encoder-Decoder (RED)\nnetwork for action anticipation. RED takes multiple history representations as\ninput and learns to anticipate a sequence of future representations. One\nsalient aspect of RED is that a reinforcement module is adopted to provide\nsequence-level supervision; the reward function is designed to encourage the\nsystem to make correct predictions as early as possible. We test RED on\nTVSeries, THUMOS-14 and TV-Human-Interaction datasets for action anticipation\nand achieve state-of-the-art performance on all datasets.","url_abs":"http://arxiv.org/abs/1707.04818v1","url_pdf":"http://arxiv.org/pdf/1707.04818v1.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":"red-reinforced-encoder-decoder-networks-for","repo_url":"https://github.com/rajskar/CS763Project","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"action-anticipation","task_name":"Action Anticipation"},{"task_slug":"decoder","task_name":"Decoder"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/action-anticipation-on-epic-kitchens-55-seen","task":"Action Anticipation","dataset":"EPIC-KITCHENS-55 (Seen test set (S1))","model":"ED","rank_in_archive_order":5,"of":7,"metrics":{"Top 1 Accuracy - Act.":"8.08","Top 1 Accuracy - Noun":"16.07","Top 1 Accuracy - Verb":"29.35","Top 5 Accuracy - Act.":"18.19","Top 5 Accuracy - Noun":"38.83","Top 5 Accuracy - Verb":"74.49"},"uses_additional_data":false},{"leaderboard":"/sota/action-anticipation-on-epic-kitchens-55-1","task":"Action Anticipation","dataset":"EPIC-KITCHENS-55 (Unseen test set (S2)","model":"ED","rank_in_archive_order":5,"of":7,"metrics":{"Top 1 Accuracy - Act.":"2.65","Top 1 Accuracy - Noun":"7.81","Top 1 Accuracy - Verb":"22.52","Top 5 Accuracy - Act.":"7.57","Top 5 Accuracy - Noun":"21.42","Top 5 Accuracy - Verb":"62.65"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1707.04818","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}