Resumen
Road markings and signs provide vehicles and pedestrians with essential information that assists them to follow the traffic regulations. Road surface markings include pedestrian crossings, directional arrows, zebra crossings, speed limit signs, other similar signs and text, and so on, which are usually painted directly onto the road surface. Road markings fulfill a variety of important functions, such as alerting drivers to the potentially hazardous road section, directing traffic, prohibiting certain actions, and slowing down. This research paper provides a summary of the Yolov5 algorithm series for road marking sign identification, which includes Yolov5s, Yolov5m, Yolov5n, Yolov5l, and Yolov5x. This study explores a wide range of contemporary object detectors, such as the ones that are used to determine the location of road marking signs. Performance metrics monitor important data, including the quantity of BFLOPS, the mean average precision (mAP), and the detection time (IoU). Our findings shows that Yolov5m is the most stable method compared to other methods with 76% precision, 86% recall, and 83% mAP during the training stage. Moreover, Yolov5m and Yolov5l achieve the highest score, mAP 87% on average in the testing stage. In addition, we have created a new dataset for road marking signs in Taiwan, called TRMSD.