Binocular vision-based road object detection and distance measurement method

With the rapid development of intelligent transportation systems and autonomous driving technology, accurately and efficiently detecting and measuring targets in complex road environments has become one of the key technologies. This paper proposes a road target detection and measurement method based on binocular vision to address the issues of target detection and distance measurement in road scenes. The binocular vision technique, by incorporating depth information, enhances the accuracy of target distance measurement. For target detection, the YOLOv7 algorithm is employed, which optimizes the loss function to effectively improve detection accuracy. Combined with the SGBM stereo matching algorithm, it enables precise distance measurement of the targets. Experimental results show that the relative measurement error is within 3%, and the model's precision reaches 89.4%, meeting the requirements for road target detection and distance measurement.