Throwing Activity Detection Using CCTV and Video Analytics for Safety and Security in Railway Station

CCTV and video analytics could assist in maintaining the safe and secure aspect of smart city. One necessity of keeping safety and security in railway stations is aggression monitoring such as throwing. A video analytics feature using video analytics and OpenPose is proposed to be added to an existing VIANA platform for railways station in Bandung. Deep learning processes to get this feature integrated in VIANA must be done including dataset preparation, training, inferencing design, and integration. This feature is proven to be effective functionally and nonfunctionally with a 76% precision, 86% recall, and 79% accuracy as well as 28% GPU utilization, 6% memory utilization and 61°C GPU temperature.