REAL TIME ROAD DAMAGE DETECTION USING YOLOV8

Abstract: Real-time road damage detection is crucial for efficient transportation infrastructure maintenance and enhanced road safety. This study applies the YOLOv8 algorithm for real-time road damage detection, aiming to provide timely and accurate information for effective decision-making and prompt repairs. The YOLOv8 algorithm was fine-tuned to identify bounding boxes associated with various types of road damage, including cracks of different sizes and irregular shapes. The model's performance was evaluated using precision, accuracy (R), F1-score, and mean average precision (mAP) indicators, demonstrating its effectiveness in detecting road damages with high accuracy. The successful implementation of real-time road damage detection systems has significant implications for proactive and targeted repairs, ultimately contributing to improved road safety and infrastructure longevity. By leveraging the power of YOLOv8 and utilizing a comprehensive dataset, this study provides an accurate, efficient, and deployable solution for real-time road damage detection. The findings have the potential to advance transportation infrastructure maintenance practices, particularly in the context of the United States, fostering safer and more sustainable road networks. Future research will focus on uploading the model to a mobile application, making it accessible to everyone and enabling widespread community participation in road condition monitoring and reporting.

Keywords: YOLOv8, cracks, precision, accuracy (R), F1-score, mean average precision (mAP), real-time road damage detection