Suspicious Activity Tracker

The ongoing exploration into detecting potentially concerning human behavior within surveillance footage remains a significant focus within the realms of image processing and computer vision. By leveraging visual surveillance methodologies, researchers aim to monitor and oversee human actions across a spectrum of environments, from sensitive to public spaces such as transportation hubs, financial institutions, shopping centers, educational institutions, parking facilities, and roadways. The primary goal is to enhance security measures, prevent criminal activities such as terrorism, theft, and vandalism, minimize accidents, and address various safety concerns. However, the challenge lies in the continuous monitoring of public areas, necessitating intelligent video surveillance systems capable of real- time human activity analysis, classification, and alert generation. Despite the increasing number of publications on visual surveillance, specifically targeting abnormal activity recognition, there remains a gap in comprehensive reviews covering diverse abnormal behaviors. This gap motivates our concise overview of suspicious human activity recognition, addressing associated challenges and methodologies. Our study focuses on six key abnormal activities: abandoned object detection, theft detection, fall detection, road accidents and illegal parking detection, violence activity detection, and fire detection. Throughout our examination, we explore various techniques outlined in the literature for identifying and analyzing human activities in surveillance videos, encompassing processes such as foreground object extraction, object detection using tracking and non-tracking methods, feature extraction, classification techniques, and subsequent activity analysis and recognition.