Machine learning models for drowsiness detection

Abstract

Road crashes and other accidents have become the common cause of fatalities and injuries in the human world. According to data from the World Health Organisation (WHO) in 2015, nearly 1.25 million people died worldwide due to road accidents. Driver fatigue is a significant factor in many road accidents. A sleepy driver is more dangerous than a driver driving at high speeds as he is victim of less sleep. Many researchers and manufactures are trying to solve this using various technologies. Driver drowsiness detection can help prevent a huge number of sleep induced road accidents. We will be using computer vision algorithms to extract facial features such as eye closure and yawning, followed by machine learning techniques to effectively detect driver state. We will be comparing multiple machine learning models and will be using the most effective one to develop a real-time drowsiness detector. This system will warn the driver if it detects a drowsy state, hence preventing any harm that may have been caused to the driver and the passengers otherwise.