Drowsy Driving has been one of the major causes behind the dangerous accidents occurring these days in the world. Drowsy Driving refers to the combination of driving and lack of sleep. In this busy world, it has become difficult to get sufficient sleep due to hectic schedules. Lack of sleep leads to sleep deprived driving which is prone to more fatal accidents compared to drunk driving. According to the Statistics, around 2,50,000 number of drivers fall asleep daily while driving. About quarter of the accidents in the world are due to sleep deprived sleeping.
One such solution which can curb the number of accidents due to drowsiness is by detecting if the driver is sleepy and alarming. This project is developed to implement one such Drowsy Driver Detection System. This system detects the eyes of the driver and alarms if the driver is sleepy. A driver is termed sleepy, if he/she closes the eyes for more than a certain period of time. Through this system, we can reduce the number of accidents in a great number thus helping the society in a significant way.
This project is implemented in two ways namely; Live detection using Computer Vision conepts and using Convolutional Neural Network to train model. Both the implementations tend to show good results in detecting the drowsiness of the driver.
Dataset used: https://sites.google.com/view/utarldd/home
Computer Vision Concept: Facial Landmarks
Neural Network architecture: Similar to VGGNet
Results and code explanation have been explained in the report.