Realization of a Safe Driving Monitoring System for Motor Vehicles

In recent years, the recognition of dangerous driving resulting from fatigue has become a cutting-edge research area. Traditional analyses of biological behavior, such as the electrical activities of the eyes and brain and the recording and analysis of driver behavioral characteristics, are prone to misjudgment, contact restriction, and low acquisition accuracy, and the combined use of these strategies must be improved. To address these problems, a safe driving monitoring system for motor vehicles is proposed in this study. Unlike contact detection, the proposed system does not require sensors and uses computer technology to achieve intelligent and accurate detection and identification. Facial features reflecting driver attentiveness are extracted from different domains by using OpenCV (Open Source Computer Vision Library) and Dlib to comprehensively assess fatigue by evaluating local features, such as the eyes, mouth, and head posture. Additionally, the anomalous item set is pretrained on the Microsoft Common Objects in Context (COCO) and UTKinect action datasets for dangerous driving recognition. Experimental results demonstrate that the proposed method has a test accuracy up to 91 %. The proposed method has high accuracy and robustness for recognizing fatigue driving and dangerous driving and can provide good fatigue warnings even when the detection of a fatigue feature is blocked.