Real time driver's eye state recognition based on deep mobile learning

Abstract

Abstract. Eye state recognition has been the subject of many studies due to its importance in many fields especially drowsy driver detection, which is crucial task that must be done in real time and mostly using limited hardware. These restrictions make resource consuming learning techniques such as deep learning difficult to use. Deep mobile learning seems to be a viable solution to solving this issue. In this paper, we propose a real time system based on deep mobile learning to classify the eye state, and compare its performance with classical machine learning methods. The experimental results on the Closed Eyes in the Wild (CEW) and MRL Eye Datasets show that the proposed approach outperformed the other machine learning techniques in terms of accuracy and execution time. In addition, we evaluated our system on a video dataset to demonstrate its reliability and robustness.