Anomaly detection and classification from CCTV camera feed

Abstract

Recently, there has been an increase in the frequency of crimes and accidents, posing a growing challenge for humans to report these incidents to the relevant authorities promptly. It is nearly impossible for humans to monitor these surveillance cameras continuously. Hence, this drawback creates a need to automate this process accurately. To address this issue, we suggest a remedy involving the utilization of CCTV feeds. Moreover, there is a need to display which frame and parts of the recording contain the anomaly, which helps to quickly judge whether that anomaly is unusual or suspicious. By incorporating concepts of convolutional neural networks (CNN) and recurrent neural networks (RNN), we aim to predict the nature of anomalies in the video footage.