Video-Computer Technology of Real Time Vehicle Driver Fatigue Monitoring

Abstract

This article is devoted to the actual problem of human fatigue control and attention concentration reduction in transport. The authors consider the most efficient, by their mind, method of person psycho-emotional condition assessment that is video control based on eye condition analysis. It is based on a convolutional neural network having its own topology. The problem of network optimal depth choosing to operate in real-time mode, and the problem of large accuracy indicators on a single-board ARM processor architecture computer were analyzed. As a research result, the software and hardware complex prototype was presented. This prototype allows to detect human fatigue by means of eye video image analysis. This system allows to reduce number of car accidents associated with vehicle driver falling asleep. In conclusion, the short-term project development prospects are proposed. Fatigue of the person who makes control, management or decision-making, and decrease of attention concentration on object can lead to critical consequences. The most efficient person physiological state control is video control based on eye condition analysis. The algorithm based on convolutional neural network and its hardware implementation, providing face search in the image, eye detection and analysis of eye condition by “open-closed” principle is proposed.