Deep learning based drivers fatigue detection in embedded system

Traffic accidents are occurred by a combination of causes such as misbehavior, carelessness, and negligence. As a result of these accidents, lethal accidents and property loss are experienced. One of the main causes of traffic accidents in the world is due to tired and sleepless driving. For this reason, the instantaneous situation of the driver in the vehicle can be monitored and fatigue can be detected and the number of accidents can be greatly reduced. For this, there is a need for a system that works in real-time, continuously monitors the driver, and can work with high accuracy. In addition, this system must be operable on an embedded device in order to be placed in the vehicle. In this study, a new approach that works in real-time on a low cost embedded device and shows high accuracy based on deep learning is proposed in order to solve the related problem. The proposed system classifies four different situations using the driver's eye and mouth areas on the Nvidia Jetson Nano embedded device. Thus, it is aimed to minimize the loss of life and property by preventing a possible accident.