Enhanced Fatigue Driving Behavior Detection Based on Iris Recognition and Facial Analysis

To more effectively identify and prevent fatigue driving behavior, this paper proposes an enhanced fatigue driving detection technique that incorporates iris recognition. This method integrates deep learning, facial feature analysis, and iris recognition technology to precisely identify and effectively combat fatigue driving. In its implementation, we use Convolutional Neural Networks (CNN) and three dimentional convolution (3D) to monitor the driver's eye, mouth, and iris status in real-time. This is done in conjunction with fatigue biological features such as blinking frequency, eyelid closure duration, and yawning frequency to determine the driver's state and categorize fatigue types. Further, we introduce an attention mechanism allowing the model to focus more on key fatigue features. As a result, the accuracy(ACC) has improved by 5%, and our algorithm can be successfully embedded into edge devices. Experimental evidence show that the proposed technique has high accuracy and reliability, making it of significant application value for preventing fatigue driving.