Recognizing driver inattention by convolutional neural networks

Driver inattention has long been recognized as the main contributing factors in traffic accidents. Development of intelligent driver assistance systems with embedded functionality of driver vigilance monitoring is therefore an urgent and challenging task. This paper presents a novel system which applies convolutional neural network to automatically learn and predict the state of driver´s eye, mouth and ear. The initial inspiration is to predict driver fatigue and distraction by analysing these states. In our works, a CNN model was trained with six classes of labeled data. The Approach was verified using self-specified Driving Dataset, which comprised of four activities, including normal driving, responding to a cell phone call, eating and falling asleep. Experiment results demonstrate that our design achieves a promising performance with a overall accuracy of 95.56% in classifying six states of the driver´s eye, mouth and ear.