Driver Drowsiness Detection

Abstract

Driver fatigue is responsible for a significant number of accidents and has emerged as a prominent factor contributing to traffic collisions. According to recent statistics, up to 30% of road accidents occur due to drowsiness. To prevent this, we need to develop a system that can detect drowsiness in real-time and alert the driver, potentially saving lives. There have been different proposed solutions to tackle this issue. Some solutions rely on physiological signals like ECG, EEG, heart rate changeability, and blood pressure, but these are not practical for real-life driving situations. Recently, computer vision-based techniques using approaches such as SVM, Naive Bayes, and other models which use deep learning have been used. The suggested method for detecting drowsiness involves utilizing a blend of computer vision methodologies and machine learning algorithms. It analyzes key features of the facial part such as the eyes and mouth, which indicates signal for drowsiness in the driver. Through continuously monitoring these regions, the system can identify signs of fatigue and alert the driver before a dangerous situation arises. The system is trained using a large dataset of labeled images, including various eye and mouth states associated with both alertness and drowsiness. Machine learning algorithms are employed to learn the patterns and features indicative of drowsiness. The accuracy of the model is continuously improved through iterative training and testing processes. The evaluation of the drowsiness detection system demonstrates its high accuracy, achieving a remarkable performance of 97% during training and 96% during testing. This accuracy metric reflects the system's ability to correctly identify drowsiness in real-world scenarios.