Analisis Deteksi Mata Kantuk Di Wajah Pengemudi Menggunakan Support Vector Machine (SVM) Berbasis Citra Real-Time

Traffic accidents in Indonesia are a serious issue with a high number of fatalities, and one of the main causes is microsleep, which is a brief moment of sleep while driving. To address this problem, this research has developed a sleePIness detection system based on the Internet of Things (IoT) using a Raspberry PI and a webcam, utilizing the Support Vector Machine (SVM) algorithm. The system is designed to detect the driver’s eye condition and provide a warning through a buzzer if the eyes are closed for more than 3 seconds. The research results indicate that the SVM model with a polynomial kernel has a training accuracy of 85.04%, demonstrating its ability to classify eye data into "opened" and "closed" categories. Evaluation with various SVM kernels, including linear, radial basis function (RBF), and polynomial, shows that the polynomial kernel performs the best with an accuracy of 85%, precision of 86%, and recall of 85% in detecting closed eyes. Although the system is effective in real-time detection of driver sleePIness, challenges remain with lighting conditions and camera positioning. Further testing is needed to improve the reliability and accuracy of the system in various situations. By providing early warnings to drivers, this system has significant potential to enhance road safety and prevent accidents caused by drowsiness.