EXPLORING DEEP LEARNING MODELS FOR SUSPICIOUS ACTIVITY DETECTION IN SURVEILLANCE FOOTAGE

Abstract

The detection of suspicious activities in surveillance footage plays a crucial role in enhancing security across various environments, including public spaces, transportation hubs, and private premises. Traditional methods of activity detection rely heavily on human operators or rule-based systems, which are often limited in their ability to handle large volumes of video data and may struggle to identify subtle or complex patterns of suspicious behavior. With the increasing availability of high-quality video surveillance data, there is a growing need for automated, intelligent systems capable of analyzing video streams in real-time to identify potential threats or 
abnormal activities.