One of the main factors contributing to traffic accidents that might cause fatalities or serious injuries is driver fatigue. The development of real-time driver sleepiness detection systems is becoming more and more popular as technology advances. In this research, we want to investigate and evaluate several approaches to sleepiness detection, with a particular emphasis on image-based metrics that employ convolutional neural networks (CNNs) and other techniques to identify fatigue in drivers. A balanced dataset will be used by the suggested system to train a CNN model. In order to correctly identify whether a motorist is attentive or sleepy, the CNN model is trained using a sizable collection of photos. When drowsiness is identified, the system can be configured to give the driver real-time alerts so they can respond appropriately. Our findings demonstrate the high accuracy with which the suggested approach can identify driver weariness, which could lead to a major decrease in the frequency of accidents resulting from driver sleepiness.