Abstract
In recent years, Smart City Infrastructures (SCI) have become familiar whereas intelligent models have been designed to improve the quality of living in smart cities. Simultaneously, anomaly detection in SCI has become a hot research topic and is widely explored to enhance the safety of pedestrians. The increasing popularity of video surveillance system and drastic increase in the amount of collected videos make the conventional physical investigation method to identify abnormal actions, a laborious process. In this background, Deep Learning (DL) models can be used in the detection of anomalies found through video surveillance systems. The current research paper develops an Internet of Things Assisted Deep Learning Enabled Anomaly Detection Technique for Smart City Infrastructures, named (IoTAD-SCI) technique. The aim of the proposed IoTAD-SCI technique is to mainly identify the existence of anomalies in smart city environment. Besides, IoTAD-SCI technique involves Deep Consensus Network (DCN) model design to detect the anomalies in input video frames. In addition, Arithmetic Optimization Algorithm (AOA) is executed to tune the hyperparameters of the DCN model. Moreover, ID3 classifier is also utilized to classify the identified objects in different classes. The experimental analysis was conducted for the proposed IoTADSCI technique upon benchmark UCSD anomaly detection dataset and the results were inspected under different measures. The simulation results infer the superiority of the proposed IoTAD-SCI technique under different metrics.