Enhanced Traffic Sign Recognition Using Advanced YOLOv8 Model

This study proposed an improved traffic sign recognition system based on the latest you only look once version 8 (YOLOv8) architecture. Efficient detection and classification of traffic signs are the keys to preventing traffic accidents and providing traffic management with the necessary control. Real-world traffic scenarios with complex lighting conditions and occlusions are usually where traditional approaches to traffic sign recognition fail to score big. Anticipating these challenges, deep learning techniques have witnessed a growth in popularity due to their enhanced accuracy and resilience in traffic sign recognition tasks. This system is designed to stand on the shoulders of the YOLOv8 model, a known leader in the area of speed and accuracy in object detection. The research focuses on classifying traffic signs into four distinct categories: traffic lights, speed limit signs, stop signs, and crosswalk markings. A dataset consisting of images of traffic signs in all environmental conditions is used for training and testing the model. The method is composed of dataset collection, data preprocessing, model architecture design, training procedure, and evaluation metrics. The accuracy and robustness of the YOLOv8 model in traffic sign recognition are measured by performance metrics like precision, recall, F1-score, and mean average precision (mAP). Comparisons of the developed approach with the state-of-the-art methods allow us to conclude the effectiveness of the approach.