Traffic Sign Detection and Recognition Using YOLOv5 and Its Versions

Abstract

A Traffic Sign Detection and Recognition (TSDR) System, which helps navigate vehicles through computer vision and human-machine communication, has to perform quickly as vehicles using them travel at high speeds. During this study, a speedy one-stage detector such as YOLOv5, a deep learning model, was chosen to dive into. This study explores creating a TSDR model by comparing four different versions of YOLOv5, namely YOLOv5 Nano, Small, Medium, and Large. This study was accomplished by first creating a new traffic sign dataset. The four versions of the YOLOv5 algorithm were then trained with a 75–25 train validation split, and 24 models were created. Afterwards, the models were tested on a test set, and their metrics were tallied. Results showed that YOLOv5 Medium and Large offer a 10% increase in accuracy performance when compared to YOLOv5 Small, but due to the slower detection speed of YOLOv5 Large, the YOLOv5 Medium models are a better fit when it comes to the detection of traffic signs when prepared by a relatively small dataset. This study provides an overview of the performance of the different YOLOv5 versions in traffic sign detection and recognition that aims to contribute to the improvement of traffic sign detectors.