Fatigue driving detection method based on deep learning

A fatigue driving detection methods based on deep learning is proposed to solve the problem of poor real-time performance and low accuracy of existing fatigue driving detection methods.Firstly, the face detection is realized by the deep learning model MTCNN.Secondly, on account of the eyes positioning is vulnerable to occlusion, posture changes and other factors, a Fine Eyes Location(FEL) model is proposed to extract the eyes region accurately and determine eyes status through OC-Net network.Finally, driver fatigue is judged based on PERCLOS algorithm and blink frequency.The experimental results show that the accuracy of the fatigue state detection achieved by this method is 97.18% and real-time requirements, and also has high robustness to complex environments.