Published Paper Details:

Driver fatigue is a significant cause of road accidents and fatalities, leading to ongoing research into methods of detecting and indicating drowsiness in drivers. Traditional approaches have included vehicle-based, behavioural-based, and physiological-based methods. However, some of these methods can be intrusive or distracting for the driver, while others require expensive sensors and complex data handling. To address these challenges, we have developed a low-cost, real-time system for actively monitoring driver drowsiness with a high degree of accuracy. Our system utilizes a webcam to record video and employs image processing techniques to detect the driver's face in each frame. Facial landmarks are identified and used to compute eye aspect ratio, mouth opening ratio, and nose length ratio. Based on these ratios and developed adaptive thresholding methods, our system can detect drowsiness in real-time. In addition, we have implemented machine learning algorithms offline to enhance the system's accuracy. Our Support Vector Machine-based classification achieved a sensitivity of 95.58% and specificity of 100%.