Abstract

This article presents the results of research, design, fabrication and application of a remote driver drowsiness system. Basing on the basic characteristics of drowsiness detection, the algorithms and software that combine the identification of features have been developed to quickly solve the problem of interference, poor lighting. Particularly training the machine according to the characteristics of the driver (face shape, nose, ears, blink time...) have made. Built on a microprocessor board that connects to external cameras, infrared LEDs, GPS / GPRS, etc., it allows on-site warnings and transmits to the monitoring center. The system allows to detect drowsiness of the driver with less than 1.5 seconds and reliability of over 90%.