A NEW ANOMALY ACTIVITY DETECTION USING CNN & RNN

Abstract

With the great use of closed-circuit tv (CCTV) surveillance structures in public areas, crowd anomaly detection has come to be an an increasing number of imperative component of the wise video surveillance system. It requires staff and non-stop interest to figure out on the captured event, which is challenging to operate via individuals. The reachable literature on human motion detection consists of a variety of procedures to observe bizarre crowd behavior, which is articulated as an outlier detection problem. This paper provides a unique evaluate of the latest improvement of anomaly detection strategies from the views of pc imaginative and prescient on exclusive handy datasets. A new taxonomic organisation of current works in crowd evaluation and anomaly detection has been introduced. A summarization of current evaluations and datasets associated to anomaly detection has been listed. It covers an overview of extraordinary crowd concepts, which include mass gathering occasions evaluation and challenges, sorts of anomalies, and surveillance systems. Additionally, lookup tendencies and future work potentialities have been analyzed.