Real-time Detection of Diverse Bangladeshi Traffic Signs Using YOLOv8

The need for intelligent traffic management solutions is crucial in the rapidly evolving fields of autonomous traffic control and navigation systems, particularly for minimizing traffic accidents and ensuring the efficient operation of traffic signals. Improving traffic safety is of paramount importance, as road accidents claim over a million lives annually worldwide, with a significant proportion of these fatalities occurring in developing countries like Bangladesh. This research aims to develop a smart assistant capable of identifying and categorizing traffic signs to enhance road safety, especially in areas where drivers may not be fully aware of traffic regulations. To achieve this, we generated a comprehensive dataset containing 5,540 distinct images of traffic signs collected from various regions across Bangladesh, including highways, rural and urban roads, and challenging locations such as steep curves. The dataset comprises 41 categories of traffic signs. The dataset includes challenges such as color variations, blurriness, occlusion, small sign sizes, and low-light conditions. To further enhance the dataset's diversity, data augmentation techniques such as random rotations, shearing, and zooming were applied. We trained the YOLOv8 deep learning model, renowned for its real-time object detection capabilities, on this dataset. The model achieved impressive results, with a mean Average Precision (mAP) of 0.80, a recall of 0.87, a precision of 0.92, and an F1 score of 0.89, demonstrating its effectiveness in real-world traffic sign detection and classification.