{"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/motion-feature-network-fixed-motion-filter","title":"Motion Feature Network: Fixed Motion Filter for Action Recognition","arxiv_id":"1807.10037","date":"2018-07-26","proceeding":"ECCV 2018 9","authors":["Myunggi Lee","Seungeui Lee","Sungjoon Son","Gyu-tae Park","Nojun Kwak"],"abstract":"Spatio-temporal representations in frame sequences play an important role in\nthe task of action recognition. Previously, a method of using optical flow as a\ntemporal information in combination with a set of RGB images that contain\nspatial information has shown great performance enhancement in the action\nrecognition tasks. However, it has an expensive computational cost and requires\ntwo-stream (RGB and optical flow) framework. In this paper, we propose MFNet\n(Motion Feature Network) containing motion blocks which make it possible to\nencode spatio-temporal information between adjacent frames in a unified network\nthat can be trained end-to-end. The motion block can be attached to any\nexisting CNN-based action recognition frameworks with only a small additional\ncost. We evaluated our network on two of the action recognition datasets\n(Jester and Something-Something) and achieved competitive performances for both\ndatasets by training the networks from scratch.","url_abs":"http://arxiv.org/abs/1807.10037v2","url_pdf":"http://arxiv.org/pdf/1807.10037v2.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":"action-recognition-in-videos","task_name":"Action Recognition"},{"task_slug":"action-recognition-in-videos-2","task_name":"Action Recognition In Videos"},{"task_slug":"optical-flow-estimation","task_name":"Optical Flow Estimation"},{"task_slug":"action-recognition","task_name":"Temporal Action Localization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/action-recognition-in-videos-on-something-1","task":"Action Recognition","dataset":"Something-Something V1","model":"Motion Feature Net","rank_in_archive_order":70,"of":74,"metrics":{"Top 1 Accuracy":"43.9"},"uses_additional_data":false},{"leaderboard":"/sota/action-recognition-in-videos-on-jester-1","task":"Action Recognition In Videos","dataset":"Jester (Gesture Recognition)","model":"MFNet","rank_in_archive_order":3,"of":9,"metrics":{"Val":"96.68"},"uses_additional_data":false},{"leaderboard":"/sota/action-recognition-in-videos-on-something-2","task":"Action Recognition In Videos","dataset":"Something-Something V1","model":"Motion Feature Net","rank_in_archive_order":2,"of":3,"metrics":{"Top 1 Accuracy":"43.9"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1807.10037","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}