Unusual Activity And Anomaly Detection In Surveillance Using GMM-KNN Model

The major purpose of surveillance is to keep track of a region for security enhancement. The traditional system is being accustomed so far, where it uses a lot of manpower and energy to manually monitor the anomalous events. It therefore results in a waste of time and resources. This paper is a part of intelligent security systems to detect any unwanted happenings or anomalies in a locale using surveillance. An approach to detect unusual activity in surveillance using GMM-KNN model has been proposed. The dataset is acquired from the industrial premises of NLC India Limited. The project proposal has the following stages. The elicitation of frames is performed from the obtained dataset. The initial foreground detection is accomplished by using the GMM. The post processing step to neglect unwanted noises and objects is done using filters and morphological operations. Statistical features are extracted for each object to detect anomaly. Feature based Detected foreground objects are classified into anomaly and background based on KNN classifier.