Identification of Driver Status Hazard Level and the System

• DOI: 10.3390/s23177536
• Publication Date: 2023-08-30T14:30:52Z
• Authors: Jiayuan Gong; Shiwei Zhou; Wenbo Ren
• Keywords: Grading (engineering); Feature (linguistics)

According to the survey statistics, most traffic accidents are caused by driver's behavior and status irregularities. Because there is no multi-level dangerous state grading system at home abroad, this paper proposes a complex for real-time detection dynamic tracking of state. The uses OpenMV as acquisition camera combined with cradle head collect current driving image in dynamically, combines YOLOX algorithm OpenPose judge detecting unsafe objects cab posture, improved Retinaface face Dlib feature-point discriminate fatigue driver. experimental results show that accuracy three driver danger levels (R1, R2, R3) obtained proposed reaches 95.8%, 94.5%, 96.3%, respectively. have specific practical significance driver-distracted warnings.