Internet-of-Things-Assisted Artificial Intelligence-Enabled Drowsiness Detection Framework

Drowsiness, drunk driving, and fatigue are major causes of car accidents and consequent deaths nowadays. In this research, we introduce a real-time, Internet-of-Things-assisted, computer vision-enabled framework for monitoring driver sleepiness based on eye aspect ratio (EAR). When the EAR ratio drops below 0.2, a warning error is generated using a text-to-voice converter. The driver is alerted via seat belt vibrators; these data are sent to a cloud server. If the driver continues to sleep or fatigues for more than 2 s after the warning mechanism, the acceleration of the car might be lowered to prevent the accident. With the help of the no infrared camera and machine learning model, images can now be seen well both during the day and at night without compromising the image quality. The efficacy of the developed model is confirmed by testing in multiple test scenarios using different subjects on a Raspberry Pi 4B graphics processing unit computer. As a result, the suggested technique has detected driver exhaustion and raised awareness and prevented the accident rate.