{"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/action-tubelet-detector-for-spatio-temporal","title":"Action Tubelet Detector for Spatio-Temporal Action Localization","arxiv_id":"1705.01861","date":"2017-05-04","proceeding":"ICCV 2017 10","authors":["Vicky Kalogeiton","Philippe Weinzaepfel","Vittorio Ferrari","Cordelia Schmid"],"abstract":"Current state-of-the-art approaches for spatio-temporal action localization\nrely on detections at the frame level that are then linked or tracked across\ntime. In this paper, we leverage the temporal continuity of videos instead of\noperating at the frame level. We propose the ACtion Tubelet detector\n(ACT-detector) that takes as input a sequence of frames and outputs tubelets,\ni.e., sequences of bounding boxes with associated scores. The same way\nstate-of-the-art object detectors rely on anchor boxes, our ACT-detector is\nbased on anchor cuboids. We build upon the SSD framework. Convolutional\nfeatures are extracted for each frame, while scores and regressions are based\non the temporal stacking of these features, thus exploiting information from a\nsequence. Our experimental results show that leveraging sequences of frames\nsignificantly improves detection performance over using individual frames. The\ngain of our tubelet detector can be explained by both more accurate scores and\nmore precise localization. Our ACT-detector outperforms the state-of-the-art\nmethods for frame-mAP and video-mAP on the J-HMDB and UCF-101 datasets, in\nparticular at high overlap thresholds.","url_abs":"http://arxiv.org/abs/1705.01861v3","url_pdf":"http://arxiv.org/pdf/1705.01861v3.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":"action-tubelet-detector-for-spatio-temporal","repo_url":"https://github.com/vkalogeiton/caffe","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"unanswered"}},{"paper_slug":"action-tubelet-detector-for-spatio-temporal","repo_url":"https://github.com/qingzhiwu/pytorch-act-detector","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"action-localization","task_name":"Action Localization"},{"task_slug":"spatio-temporal-action-localization","task_name":"Spatio-Temporal Action Localization"},{"task_slug":"action-recognition","task_name":"Temporal Action Localization"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"non-maximum-suppression","method_name":"Non Maximum Suppression"},{"method_slug":"ssd","method_name":"SSD"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1705.01861","atlas_url":"https://app.syntology.ai/?focus=1705.01861","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}