Real-time Driver Drowsiness Detection System using Cascaded ConvNet Framework

Driver fatigue and drowsiness are major causes of traffic accidents. Globally, the number of casualties and fatalities increases each year. These accidents cause a lot of serious and fatal injuries each year, which is very expensive for both individuals and governments. Many drivers suffer from drowsiness as a result of their long workdays, insufficient sleep, and fatigue. Due to their reduced concentration, awareness, and control over their vehicles, tired drivers are just as dangerous as drunk ones. The goal of this research study is to provide an overview of an ADAS (Advanced Driving (Assistance System) that is specifically designed to detect driver fatigue and avert traffic accidents. This study has created a machine-learning algorithm to solve this problem. Upon detecting pupil and eye movement, the devices (after model installation) can make an audible noise to keep the drivers awake. The proposed model was tested on the Inception v3 model. The proposed model is trained using a convolutional neural network (ConvNet) and transfer learning, and it achieved 98.1% accuracy.