An approach for traffic sign recognition

Abstract

This article presents a model for detecting and recognizing traffic signs based on the YOLO (You Only Look Once) algorithm. Our system can detect traffic signs in real-world scenarios, including prohibitory, stop, no entry, speed limit, regulatory, and hazardous signs. However, there are still some cases where successful recognition is not achieved. Experiments were conducted on a dataset of 29,632 images, yielding % recognition accuracy of 86.8%. The system performs well in practical environments with relatively high accuracy, yet some errors persist during detection.