Anomaly Detection in Video Using Gaussian Model and Recurrent Neural Network
P Kola Sujatha, S Yuvarani · 2018
Anomaly detection in surveillance system demands to address various problems. The proposed work aims to identify different types of anomalies in different datasets such that traffic anomaly in QMUL, a pedestrian anomaly in Avenue and loitering anomaly Subway. Gaussian Kernel Integration Model (GKIM) has been used for extracting features. Classification of anomaly has been performed by using RCRF. Here spatial and temporal features integrated into Gaussian kernel model. Existing work was carried out for pedestrian anomaly only. Proposed work says that GKIM model works well in different types of anomaly. Performance analysis has been carried out in terms of Equal Error Rate and Detection Rate.