Practical research on lane recognition and driving state monitoring method based on computer vision

This paper mainly discusses the monitoring method of lane recognition analysis and real-time monitoring of driver driving status by vehicle-mounted camera using computer vision technology, and carries out practical research by combining the monitoring method with specific application scenarios. Driver face recognition and road straight line detection are implemented through OpenCV to provide certain auxiliary functions for driver driving. By using Yolov5 for fatigue detection and detection of objects such as cell phones and water cups, in order to improve driver concentration and enhance driving safety. More and more cars on the market are being equipped with driver assistance systems. Some research data shows that driver assistance systems can increase driving safety, but the high selling price reduces the audience. Cars belong to the category of durable goods, and there are still a large number of ordinary cars driving on the road. We have developed a low-cost, general vehicle-specific solution to this problem. This product does not require expensive LIDAR sensors and other equipment and does not require significant changes to the vehicle in terms of deployment. The software and hardware are low cost and cater to the market demand. The latest technologies and frameworks such as OpenCV and Yolov5 are also used to develop face recognition and detection, item recognition, and blink count through training datasets.