Comprehensive Survey on Object Detection and Anomaly Detection Techniques for Ensuring Pedestrian Safety in Walkways

Ensuring pedestrian safety in walkways gains more significance in urban surroundings. Anomaly detection (AD) approaches had a crucial role in assuring pedestrian safety in pathways, where abnormal events or behaviours can give rise to potential risks. To attain this, potential AD and object detection approaches are essential to identify abnormal behaviours and potential hazards in real-time. This survey presents a complete overview of AD and object detection algorithms particularly devised for pedestrian walkways. The survey starts by discussing the importance of pedestrian safety and the difficulties related to identifying anomalies and objects in walkway settings. Then it delivers a comprehensive analysis of different object detection algorithms that includes existing deep learning methods and classical computer vision (CV) based methods. The applications, strengths, and limitations of all methods are inspected, emphasizing their potentiality in pedestrian tracking and detection. Moreover, the survey explored AD approach that intends to determine abnormal events or behaviours in walkways. Such methods have machine learning-based methods, statistical modeling, and AD methods designed for pedestrian-centric situations. The survey inspects their respective pros, cons, and real-time applications. The survey concluded by discussing future directions and emerging trends in object detection and AD for pedestrian safety in walkways.