Automating the Detection of Road Damage via the Use of UAV-Captured Images and the YOLO

This research introduces an interesting methodology for the automated recognition of road damage employing images captured by Unmanned Aerial Vehicles (UAVs) and the YOLO (You Only Look Once) Deep Learning (DL) algorithm. Although conventional road inspection techniques are risky and wasteful, maintaining a reliable road network is essential to public safety and economic growth. This work utilizes UAVs to capture high-resolution images of road surfaces, which are subsequently analyzed using YOLOv5, YOLOv7, and the latest YOLOv8 models. Through comprehensive training and testing, including data augmentation techniques, models in this work are fine-tuned for high accuracy in road damage detection. The results demonstrate that YOLOv8 achieves the highest accuracy, significantly outperforming earlier versions. This study underscores the ability to integrate UAV technology with DL for efficient, accurate, and scalable road maintenance solutions.