Fatigue Driving Recognition Based on Deep Learning

The increasing number of motor vehicles in China brings convenience to people ' s lives. At the same time, the incidence of road traffic accidents and casualties have increased year by year, which poses a serious threat to society and individuals. Scholars at home and abroad pay great attention to the research of fatigue driving detection, and have made more research results. The traditional target detection method based on computer vision is difficult to be widely used due to its poor adaptability, high computational cost and poor real-time effect. To solve this problem, this paper proposes a fatigue driving detection module based on deep learning, which locates human eyes, mouth and head by detecting the model detecting the model. At the same time, the robustness of the recognition model is improved by combining fine-grained classification, data enhancement and other methods, and the semi-closed state of eyes, mouth and head posture are accurately identified. Then, the recognition state of each part is combined with the PERCLOS criterion. Finally, the experiment proves that the model can make accurate fatigue judgment for the driver through accuracy and recall rate, showing its excellent performance.