Traffic Sign Recognition using YOLOv4

Traffic sign recognition can provide road information assessment and real-time safety early warning for safe driving of vehicles. The YOLOv4 model was adopted for traffic sign detection and recognition, and the K-means clustering algorithm is used to improve YOLOv4. A test was performed with the CCTSDB dataset. We achieved a 25.3fps and 90.42% mAP on the GTSDB dataset.