Description: The problem addressed is the effective analysis of vast volumes of surveillance data to detect potential threats or anomalies in crowd behavior. To tackle this challenge, a novel framework called ‘CRAB-NET’ has been developed for automated behavior recognition using Convolutional Long Short-Term Memory networks (ConvLSTM) and Long-Term Recurrent Convolutional Networks (LRCN). Additionally, a significant contribution of this research is the creation of a diverse and representative video dataset accurately reflecting real-world crowd dynamics across 8 different categories.
Socio-Economic Benefits: The socio-economic benefit of this project is the enhancement of public safety and security during mass gatherings and events through the proactive identification of potential threats using the ‘CRAB-NET’ framework. This approach enables timely interventions to prevent casualties, maintain order, and allocate security resources more efficiently, thereby reducing economic losses associated with disruptions or incidents.
Category: Computing
Department: CSSE