This paper presents a novel approach to detecting signs of driver drowsiness by focusing on human facial characteristics such as eye closing length and eye flickering. The proposed model effectively addresses the limitations of previous methods, offering a more accurate and reliable means of detecting driver drowsiness, unlike existing techniques that rely on physiological parameters (like the direct placement of electrodes on the driver's body) or algorithms like Viola-Jones. This model utilizes Convolutional Neural Network (CNN) and Multi-Channel Kernelized Correlation Filter (MC-KCF) to analyze the features for facial recognition. The use of CNN allows the system to process both videos and pictures, providing a comprehensive analysis of the driver's facial expressions. By training the model on a large dataset, the proposed model achieves an impressive accuracy of approximately 97%.