DEEP CNN BASED APPROACH FOR DRIVER DROWSINESS DETECTION

Driver sleepiness is one of the leading causes of road accidents. Drowsiness is the main risk factor that causes traffic crashes, several injuries, and a high risk of fatalities. Deep learning has made some progress in identifying the driver drowsiness while driving a vehicle. In this study, we propose a two-dimensional CNN-based classification model to extract the information from facial images and categorize it into sleepy and non-sleepy classes. The performance of the model was compared to that of other transfer learning techniques, such as VGG-16 and ResNet-50. Furthermore, the validation accuracy of each model has been evaluated and measured along with precision, recall, and f1-score. According to the evaluations, the proposed model exhibits better performance than other transfer learning strategies.