VIGILANT – A Car Accident Prevention System based on Driver Drowsiness

The proposed system namely Vigilant detects the driver’s drowsiness by monitoring the eye and the mouth movements. This system precisely detects unusual blinking of eyes and mouth movements commonly known as yawning. A pre-trained model which uses facial landmarks algorithm is used to detect the eye and mouth movements. After detecting the eye and mouth movements drowsiness is calculated based on predetermined thresholds. The threshold values are set based on real time situation of driver. All the data required in the process is collected and updated on a firebase cloud service. When the drowsiness is detected, an in-vehicle buzzer alarm is triggered. SOS signal is sent to an emergency contact number. Also, the respective data can be notified to hospital and police authorities through a common portal. The data includes live location, eye blink count, yawning count, ride start time, amount of drowsiness. Even when no drowsiness is detected all the data is updated to the cloud and can be accessed by the respective user only. This system is implemented using a Raspberry Pi Board. Various sensors are integrated to determine parameters used in decision making. The system has high accuracy low latency and works well under low lighting conditions.