Drowsiness/Sleep Detection and Alert System: Development of A Prototype Using Raspberry PI

Road traffic accidents are a critical issue because they cause death and disability to Thai people. This paper presents the pilot study of the proposed integrated drowsiness/sleep detection and alert system to reduce the rate of road accidents in Thailand that is very high compared to other countries. In this pilot phase, the detection system has been developed using both hardware (Raspberry PI 3, a camera and a loudspeaker) and the software (Open CV, Python and Dlib). The system will detect from the video data obtained from a camera to check whether a driver's eyes are closed for longer than the pre-set time or not. If so, the buzzer will sound an alarm. After testing and evaluation by 3 and 35 subjects, respectively, it has been found that its average performance is 98.15 %, while the overall score for user satisfaction is $4.66 ± 0.28$. The main issue of this system is the performance of detection at night. Therefore, it should be improved in the next phase that will be enhanced with additional parts and features.