Research and Implementation of the Key Technology of the Intelligent Vehicle Occupant Monitoring System

With the rapid development of intelligent vehicles, occupant safety and interactive experience have become a research hotspot. This study focuses on the key technologies in the intelligent vehicle occupant system (OMS), and proposes the visual perception scheme based on OpenPose and Mediapipe, aiming to solve the problem of human-computer interaction efficiency between the child safety risks and the rear passengers. The OpenPose algorithm detects the key points of passenger bones in real time, and calculates the actual sitting height and shoulder width. With the Mediapipe hand key point detection and LSTM model, the rear passenger gesture recognition system is designed to control the on-board multimedia function. The experimental results showed that the detection accuracy of the passenger children was 92.38 %, the accuracy of gesture recognition was 92.83 %, and the average response time was 12 frames / s. The system effectively improves the passenger safety and interactive experience, and provides innovative solutions for the intelligent cockpit technology.