Traffic Signs Detection and Recognition System Using the YOLOv4 Algorithm

Traffic signs are one of the important road equipment facilities to inform road users about regulations and visual directions. Currently, an automatic Traffic Sign Recognition (TSR) system is being developed which is implemented in an advanced driver system (ADAS) so that road users can be safe and secure while on the road. Therefore, this paper aims to be able to detect and recognize traffic signs on the highway to provide information on the meaning of these traffic signs automatically. In this study, 35 classes of signs were used which consisted of warning signs, prohibitions signs, mandatory signs, and instructions signs. This system is implemented using the Darknet framework with the You Only Look Once version 4 (YOLOv4) model. The investigation carried out in this study is a system that detects and recognizes traffic signs evaluated on offline-based video in one-way traffic during the day. The result of mAP (mean Average Precision) in this system is 95.15%.