Deep Anomaly Detection Approaches for Enhancing Pedestrian Safety

Anomaly detection in pedestrian walkways is a critical research area to enhance pedestrian safety. The overspeeding of vehicles, drunk driving, bad weather, and failures in autonomous vehicles pose a threat to pedestrians. The imperative for a smart surveillance system capable of autonomously detecting abnormal activities is paramount. In this scenario, this paper presents a deep learning-based anomaly detection system for enhancing pedestrian safety. The proposed approach analyzes the data captured by the IP cameras to identify the movement of vehicles on pedestrian walkways and sends text alert messages. The deep anomaly detection approaches presented in this research study integrate Convolutional Autoencoder and Fuzzy Logic with Data Encryption techniques for developing a context-sensitive smart surveillance model. The model has been trained with 1200 images consisting of pedestrian paths which are taken from the University of California San Diego (UCSD) dataset. The proposed anomaly detection systems’ effectiveness is assessed using metrics such as accuracy, precision, recall, F1-score, and area under curve (AUC).