Dress Identification for Camp Security

Abstract

With the development of artificial intelligence and internet of things, the intelligent embedded devices are becoming more and more popular. So it is urgent to achieve the lightweight application of artificial intelligence. In this paper, a dress identification framework is developed for camp security. The framework has both hardware and software. The hardware is composed of LattePanda board, Arduino chip, USB camera, and buzzer. To achieve the identification, we extract the color histogram feature of moving person from the images captured by USB camera, and identify the military dress by using support vector machine algorithm. The equipment can output sound or light signals by Arduino chip, when it identifies a non-military dress. We implement the identification function based on the OpenCV library. The framework can run in real-time, with a reliable precision.