Aiming at the problems such as the difficulty of measuring the angle of traffic sign installation and the single measurement method. Through YOLOv8 object detection accuracy experiments, the model is made to have an accurate object detection frame to ensure the accuracy of signage pixel coordinates. In this paper, iterative geometric encoder stereo matching model (IGEV) is introduced to improve the binocular ranging model using migration learning. Ranging experiments were set up in 1.5m, 2.0m, 2.5m and 3m four distance variables, after the experiments to verify the accuracy of the ranging model, and the true distance value compared with the ranging error rate of 0.67%; sign installation angle experiments proved that all the error rate of the angular measurement is less than 7.62%. This paper method sign installation angle measurement method intelligent.