{"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/future-frame-prediction-for-anomaly-detection","title":"Future Frame Prediction for Anomaly Detection -- A New Baseline","arxiv_id":"1712.09867","date":"2017-12-28","proceeding":null,"authors":["Wen Liu","Weixin Luo","Dongze Lian","Shenghua Gao"],"abstract":"Anomaly detection in videos refers to the identification of events that do\nnot conform to expected behavior. However, almost all existing methods tackle\nthe problem by minimizing the reconstruction errors of training data, which\ncannot guarantee a larger reconstruction error for an abnormal event. In this\npaper, we propose to tackle the anomaly detection problem within a video\nprediction framework. To the best of our knowledge, this is the first work that\nleverages the difference between a predicted future frame and its ground truth\nto detect an abnormal event. To predict a future frame with higher quality for\nnormal events, other than the commonly used appearance (spatial) constraints on\nintensity and gradient, we also introduce a motion (temporal) constraint in\nvideo prediction by enforcing the optical flow between predicted frames and\nground truth frames to be consistent, and this is the first work that\nintroduces a temporal constraint into the video prediction task. Such spatial\nand motion constraints facilitate the future frame prediction for normal\nevents, and consequently facilitate to identify those abnormal events that do\nnot conform the expectation. Extensive experiments on both a toy dataset and\nsome publicly available datasets validate the effectiveness of our method in\nterms of robustness to the uncertainty in normal events and the sensitivity to\nabnormal events.","url_abs":"http://arxiv.org/abs/1712.09867v3","url_pdf":"http://arxiv.org/pdf/1712.09867v3.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":"future-frame-prediction-for-anomaly-detection","repo_url":"https://github.com/stevenliuwen/ano_pred_cvpr2018","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"anomaly-detection","task_name":"Anomaly Detection"},{"task_slug":"optical-flow-estimation","task_name":"Optical Flow Estimation"},{"task_slug":"prediction","task_name":"Prediction"},{"task_slug":"video-anomaly-detection","task_name":"Video Anomaly Detection"},{"task_slug":"video-prediction","task_name":"Video Prediction"}],"methods":[],"datasets_introduced":[{"slug":"shanghaitech-campus","name":"ShanghaiTech Campus","full_name":""}],"methods_introduced":[],"results":[{"leaderboard":"/sota/anomaly-detection-on-chuk-avenue","task":"Anomaly Detection","dataset":"CUHK Avenue","model":"Future Frame Prediction","rank_in_archive_order":31,"of":35,"metrics":{"AUC":"85.1%","FPS":"25","RBDC":"19.59","TBDC":"56.01"},"uses_additional_data":false},{"leaderboard":"/sota/traffic-accident-detection-on-a3d","task":"Traffic Accident Detection","dataset":"A3D","model":"State-of-the-art","rank_in_archive_order":3,"of":3,"metrics":{"AUC":"46.1"},"uses_additional_data":false},{"leaderboard":"/sota/traffic-accident-detection-on-sa","task":"Traffic Accident Detection","dataset":"SA","model":"State-of-the-art","rank_in_archive_order":3,"of":3,"metrics":{"AUC":"50.4"},"uses_additional_data":false},{"leaderboard":"/sota/video-anomaly-detection-on-hr-avenue","task":"Video Anomaly Detection","dataset":"HR-Avenue","model":"Pred","rank_in_archive_order":8,"of":11,"metrics":{"AUC":"86.2"},"uses_additional_data":false},{"leaderboard":"/sota/video-anomaly-detection-on-hr-shanghaitech","task":"Video Anomaly Detection","dataset":"HR-ShanghaiTech","model":"Pred","rank_in_archive_order":13,"of":14,"metrics":{"AUC":"72.7"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1712.09867","atlas_url":"https://app.syntology.ai/?focus=1712.09867","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}