Recognizing Road Surface Traffic Signs Based on Yolo Models Considering Image Flips

In recent years, there have been significant advances in deep learning and road marking recognition due to machine artificial intelligence. Despite progress, it often relies heavily on unrepresentative datasets limited situations. Drivers advanced driver assistance systems rely markings help them better understand their environment the street. Road are signs texts painted surface, including directional arrows, pedestrian crossings, speed limit signs, zebra other equivalent texts. Pavement also known as markings. Our experiments briefly discuss convolutional neural network (CNN)-based object detection algorithms, specifically for Yolo V2, V3, V4, V4-tiny. our experiments, we built Taiwan Marking Sign Dataset (TRMSD) made a public dataset so researchers could use it. Further, train model distinguish left right objects into separate classes. Furthermore, V4 V4-tiny results can benefit from “No Flip” setting. case, want The best experiment is (No Flip), with test accuracy of 95.43% an IoU 66.12%. this study, (without flipping) outperforms state-of-the-art schemes, achieving 81.22% training 95.34% testing TRMSD dataset.