DESIGN OF A TRAFFIC SIGN RECOGNITION SYSTEM WITH THE YOLO APPLICATION

Abstract

A safer and more reliable traffic sign recognition system is a need of drivers and is also a research hot spot for automobile manufacturers today. To address the above need, the authors propose in this article a traffic sign recognition system based on computer vision techniques and the You Only Look Once (YOLO) algorithm. The system is designed to recognize 9 different types of traffic signs including: No Left Turn, No Right Turn, No Left Turn and No Right Turn, No Stopping and Parking, No Parking, No Automobiles turning right, No Automobiles turning left ,No U turns , and No Going straight- ahead. To train the proposed system, images were collected from roads in Ho Chi Minh City including 735 signs with name No Left Turn, 713 signs with name No Right Turn, 177 images with No Left Turn and No Right Turn, 752 images with name No Stopping and Parking, 629 signs with name No Parking, 191 signs of prohibiting cars from turning right, 143 signs of prohibiting cars from turning left, 171 signs with name No U turns  and 109 signs of  No Going straight- ahead. Finally, the system is experimentally tested in the field for recognition accuracy according to the mAP@.5 measure.