{"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/optical-flow-guided-feature-a-fast-and-robust","title":"Optical Flow Guided Feature: A Fast and Robust Motion Representation for Video Action Recognition","arxiv_id":"1711.11152","date":"2017-11-29","proceeding":"CVPR 2018 6","authors":["Shuyang Sun","Zhanghui Kuang","Wanli Ouyang","Lu Sheng","Wei zhang"],"abstract":"Motion representation plays a vital role in human action recognition in\nvideos. In this study, we introduce a novel compact motion representation for\nvideo action recognition, named Optical Flow guided Feature (OFF), which\nenables the network to distill temporal information through a fast and robust\napproach. The OFF is derived from the definition of optical flow and is\northogonal to the optical flow. The derivation also provides theoretical\nsupport for using the difference between two frames. By directly calculating\npixel-wise spatiotemporal gradients of the deep feature maps, the OFF could be\nembedded in any existing CNN based video action recognition framework with only\na slight additional cost. It enables the CNN to extract spatiotemporal\ninformation, especially the temporal information between frames simultaneously.\nThis simple but powerful idea is validated by experimental results. The network\nwith OFF fed only by RGB inputs achieves a competitive accuracy of 93.3% on\nUCF-101, which is comparable with the result obtained by two streams (RGB and\noptical flow), but is 15 times faster in speed. Experimental results also show\nthat OFF is complementary to other motion modalities such as optical flow. When\nthe proposed method is plugged into the state-of-the-art video action\nrecognition framework, it has 96:0% and 74:2% accuracy on UCF-101 and HMDB-51\nrespectively. The code for this project is available at\nhttps://github.com/kevin-ssy/Optical-Flow-Guided-Feature.","url_abs":"http://arxiv.org/abs/1711.11152v2","url_pdf":"http://arxiv.org/pdf/1711.11152v2.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":"optical-flow-guided-feature-a-fast-and-robust","repo_url":"https://github.com/kevin-ssy/Optical-Flow-Guided-Feature","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"unanswered"}}],"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-hmdb-51","task":"Action Recognition","dataset":"HMDB-51","model":"Optical Flow Guided Feature","rank_in_archive_order":45,"of":77,"metrics":{"Average accuracy of 3 splits":"74.2"},"uses_additional_data":false},{"leaderboard":"/sota/action-recognition-in-videos-on-ucf101","task":"Action Recognition","dataset":"UCF101","model":"Optical Flow Guided Feature","rank_in_archive_order":40,"of":91,"metrics":{"3-fold Accuracy":"96"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.11152","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}