YOLO-Based Real-Time Monitoring System Using Radar And Cameras

Abstract

Recently, object recognition technology is positioned as a core technology in the overall shipbuilding industry, such as video surveillance, face recognition, robot control, autonomous driving, smart factory, and security, due to the advancement of hardware performance and miniaturization and related technologies. In this paper, we intend to develop a deep learning-based real-time monitoring system using data collected from radar and cameras. The camera is moved to the corresponding location using the location information of the object detected by the radar. In addition, the AI analysis client recognizes objects present in the image at the corresponding location and transmits the result to the server. Finally, the server determines whether there is a threat according to the object information received from the AI analysis module and sounds a warning sound. In order to develop such a system, first, an AI analysis module development environment for object recognition in an image is established, and a protocol necessary to communicate data between a server and an AI analysis module is defined. And the method to improve the real-time and accuracy of the AI analysis module applies.