2019/08/08 by Thomas Golda, Nils Murzyn, Golda, Thomas +5
Computer Science · #Anomaly Detection Techniques and Applications #Computer Vision and Pattern Recognition (cs.CV) #Digital Media Forensic Detection #FOS: Computer and information sciences #Human Pose and Action Recognition
paper · pdf · doi:10.48550/arxiv.1908.03055
openalex publication_date 2019/08/08 · openalex created_date 2022/07/23 · openalex updated_date 2026/07/28
Anomaly detection plays in many fields of research, along with the strongly\nrelated task of outlier detection, a very important role. Especially within the\ncontext of the automated analysis of video material recorded by surveillance\ncameras, abnormal situations can be of very different nature. For this purpose\nthis work investigates Generative-Adversarial-Network-based methods (GAN) for\nanomaly detection related to surveillance applications. The focus is on the\nusage of static camera setups, since this kind of camera is one of the most\noften used and belongs to the lower price segment. In order to address this\ntask, multiple subtasks are evaluated, including the influence of existing\noptical flow methods for the incorporation of short-term temporal information,\ndifferent forms of network setups and losses for GANs, and the use of\nmorphological operations for further performance improvement. With these\nextension we achieved up to 2.4% better results. Furthermore, the final method\nreduced the anomaly detection error for GAN-based methods by about 42.8%.\n