Habib Olaniyi Aliu & Joy Oluwabukola Olayiwola

One of the main contributing factors to traffic accidents is driver drowsiness, which is mostly brought on by exhaustion. Even though drowsiness only lasts a short while, it has such a detrimental effect on everyone—not just the driver. In this study, a CNN lightweight model called MnasNet is employed to identify driver drowsiness. MnasNet has the capacity to be both quick and precise. The study used a unique dataset with 6,000 photos that was then divided into four categories: yawn, no yawn, eye closed, and eye closed, respectively. A ratio of 80:10:10 was used to split the data samples for each class into training, validation, and testing groups accordingly. The model's implementation accuracy was 83.16%. Keywords: CNN, MnasNet, driver drowsiness, Machine Learning 0150