Advancing Road Safety: A Comparative Analysis of YOLO v8 an v5 for Real-Time Traffic Sign Detection and Recommendation

In the era of automation, road safety remains a significant challenge despite technological advancements. This study proposes a real-time Traffic Sign Detection and Recommendation System (TSDRS) utilizing deep learning models to enhance driver awareness and traffic management. The research primarily focuses on evaluating the performance of YOLOv8 and YOLOv5 for traffic sign detection while also incorporating Faster R-CNN and SSD for comparison. Using a custom dataset of annotated traffic sign images, key performance metrics such as mAP, precision, recall, and F1-score are analyzed to determine model efficiency. The results demonstrate that YOLOv8 outperforms YOLOv5 across all categories, achieving superior accuracy, recall, and inference speed. Additionally, a hybrid approach integrating YOLOv8 with Mask R-CNN is explored, showing improved segmentation but increased computational cost. The study also investigates the impact of data augmentation during training, revealing that augmentation enhances overall detection performance. The findings support the adoption of YOLOv8 for real-time traffic sign detection, offering a scalable solution for intelligent transportation systems and autonomous driving applications.