Drowsiness and Emotion Detection of Drivers for Improved Road Safety

Often resulting in fatalities and serious injuries, drowsy driving is a key contributing factor in collisions. Drivers who are irritable or disturbed are more likely to be involved in accidents. Identifying symptoms of exhaustion and negative emotions may facilitate taking preventative measures before an accident occurs. By identifying and responding to the driver's state, cars equipped with sophisticated technology such as drowsiness and emotion detection aim to enhance the driving experience. With the use of machine learning algorithms, these cars can accurately read facial expressions to determine the driver's drowsiness and emotion. The technology may alter the vehicle's behaviour and user interface in response to the driver's state, making driving safer and more personalised. To detect driver's drowsiness different approaches are taken in this experiment such as the use of Basic CNN, ResNet50, VGG16, VGG19 and InceptionV3. Among them InceptionV3 has given 100% accuracy in the process of detecting drowsy drivers for yawn eye dataset new. In case of NITYMED dataset where video data were used, Resnet50 produced 100% accuracy. We have used InceptionV3 for detecting driver's emotion for FER2013 Dataset, which has given 68.02% accuracy. We conducted a survey on drivers regarding what affects them at the time of driving and whether a continuous monitoring system will help them or not. Almost everyone agreed on having a monitoring system for road safety.