Enhanced Driver Drowsiness Detection Using Deep Learning: Comparative Analysis of CNN Architectures for Video-Based Monitoring Systems

This research work presents a video-based driver drowsiness detection system using deep learning models. The processed video data sets were analyzed with several CNN architectures: InceptionV3, MobileNet, DenseNet, and AlexNet. The results of the research work show that the proposed methodology obtains great accuracy in detecting the drowsiness for a driver, providing a reliable approach to minimize accidents. Its inceptionV3 network has outstanding results with an astoundingly fantastic score of about 98% in the data of a video set as well. At the close ranking, there are the superior capabilities of the performance of the mobileNet model while achieving about 97%. A study will elucidate the effectiveness of video monitoring systems based on the application of the state-of-the-art developments in deep learning to the vehicle’s video streaming system. These results indicate the advanced capabilities of modern CNN architectures in processing and interpreting complex video data, further avenues for innovation and improvement within the field.