{"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/pages-iai-uni-bonn-de-https-pages-iai-uni","title":"Intention-based Long-Term Human Motion Anticipation","arxiv_id":null,"date":"2021-12-01","proceeding":"3DV 2021 12","authors":["Julian Tanke","Chintan Zaveri","Juergen Gall"],"abstract":"Recently, a few works have been proposed to model the\r\nuncertainty of the future human motion. These works do\r\nnot forecast a single sequence but multiple sequences for\r\nthe same observation. While these works focused on in-\r\ncreasing the diversity, this work focuses on keeping a high\r\nquality of the forecast sequences even for very long time\r\nhorizons of up to 30 seconds. In order to achieve this goal,\r\nwe propose to forecast the intention of the person ahead of\r\ntime. This has the advantage that the generated human mo-\r\ntion remains goal oriented and that the motion transitions\r\nbetween two actions are smooth and highly realistic. We\r\nfurthermore propose a new quality score for evaluation that\r\ncorrelates better with human perception than other metrics.\r\nThe results and a user study show that our approach fore-\r\ncasts multiple sequences that are more plausible compared\r\nto the state-of-the-art.","url_abs":"https://pages.iai.uni-bonn.de/gall_juergen/download/jgall_forecastintention_3dv21.pdf","url_pdf":"https://pages.iai.uni-bonn.de/gall_juergen/download/jgall_forecastintention_3dv21.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":"pages-iai-uni-bonn-de-https-pages-iai-uni","repo_url":"https://github.com/zaverichintan/Intention-based-Long-Term-Human-Motion-Anticipation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"diversity","task_name":"Diversity"},{"task_slug":"human-pose-forecasting","task_name":"Human Pose Forecasting"}],"methods":[{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/human-pose-forecasting-on-human36m","task":"Human Pose Forecasting","dataset":"Human3.6M","model":"Forecast LSTM","rank_in_archive_order":17,"of":33,"metrics":{"APD":"3070","Average MPJPE (mm) @ 1000 ms":"149.2","Average MPJPE (mm) @ 400ms":"80.8","Average NDMS at 4s ":"0.465","MAR, walking, 400ms":"0.91"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}