If the driver can discover the safety hazards during the driving and brake the vehicle in time, it will effectively reduce the occurrence of traffic accidents and reduce losses. In order to accurately and quickly detect the vehicle in front, this article proposes a car target detection method based on the YOLOv7-tiny algorithm. This method can use the machine vision to detect the vehicle in front. Provide security assistance. The experimental results show that the average accuracy rate of the methods proposed in this article on six common vehicle recognition reached 80.8%, and the model is more lighter.