{"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/long-term-temporal-convolutions-for-action","title":"Long-term Temporal Convolutions for Action Recognition","arxiv_id":"1604.04494","date":"2016-04-15","proceeding":null,"authors":["Gül Varol","Ivan Laptev","Cordelia Schmid"],"abstract":"Typical human actions last several seconds and exhibit characteristic\nspatio-temporal structure. Recent methods attempt to capture this structure and\nlearn action representations with convolutional neural networks. Such\nrepresentations, however, are typically learned at the level of a few video\nframes failing to model actions at their full temporal extent. In this work we\nlearn video representations using neural networks with long-term temporal\nconvolutions (LTC). We demonstrate that LTC-CNN models with increased temporal\nextents improve the accuracy of action recognition. We also study the impact of\ndifferent low-level representations, such as raw values of video pixels and\noptical flow vector fields and demonstrate the importance of high-quality\noptical flow estimation for learning accurate action models. We report\nstate-of-the-art results on two challenging benchmarks for human action\nrecognition UCF101 (92.7%) and HMDB51 (67.2%).","url_abs":"http://arxiv.org/abs/1604.04494v2","url_pdf":"http://arxiv.org/pdf/1604.04494v2.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":"long-term-temporal-convolutions-for-action","repo_url":"https://github.com/gulvarol/ltc","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"action-recognition-in-videos","task_name":"Action Recognition"},{"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":"LTC","rank_in_archive_order":64,"of":77,"metrics":{"Average accuracy of 3 splits":"64.8"},"uses_additional_data":false},{"leaderboard":"/sota/action-recognition-in-videos-on-ucf101","task":"Action Recognition","dataset":"UCF101","model":"LTC","rank_in_archive_order":66,"of":91,"metrics":{"3-fold Accuracy":"91.7"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1604.04494","atlas_url":"https://app.syntology.ai/?focus=1604.04494","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}