Driver fatigue detection using image sensors

Abstract

Road traffic accidents happen every day in every part of the world. There are many causes for road traffic accidents. One of the causes is driver’s fatigue. In fact, fatigue driving has been one of the major causes for traffic accidents and it accounts for 20% of all traffic accidents in the world, which leads to many deaths and injuries [1]. Much research and inventions have been introduced to tackle the problem. Driver face detection and monitoring systems is one of the many solutions to fatigue driving. They can detect facial features such as eyes, nose and mouth. Apart from that, the system can extract the symptoms of fatigue. These symptoms are mainly percentage of eyelid closure over time (PERCLOS), rate of eye blink, eye sporadic movement, yawning and head nodding [2]. The robust technology that we have today for facial detections have been assisting us to analyse and recognise the symptoms in real time. This paper presents an introduction, project aim and methodology used to recognize and detect the facial features and fatigue symptoms. The general structure of the existing system will be discussed. The system will be programmed in Python. It uses the multi-task convolutional neural network (MTCNN) whereby multiple convolutional neural networks are tied together. A possible improvement will be further elaborated as well.