Road Pothole Damage Detector at Night Using YOLOv8

This study aimed to implement YOLOv8 Ultralytics to detect road potholes at night and analyze its accuracy under various lighting conditions. Detecting road potholes at night is crucial, because limited visibility increases the risk of accidents. Using data collected through web scraping and open sources, the model was trained on 3,685 images divided into 70% for training, 20% for testing, and 10% for validation. The results showed that the first scenario provided the best performance, with a precision of 83.0%, recall of 87.9%, and mAP50 of 89.9%. This indicates that the model can accurately detect more potholes. With an accuracy of 74.5% and an F1-Score of 85.4%, the first scenario offers an optimal balance between detection speed and precision, which is crucial in low-light conditions, such as at night.The main contributions of this study include the application of YOLOv8 for nighttime road pothole detection and the development of a real-time road-monitoring system to improve traffic safety. In the future, improvements can be made by expanding the data variation, such as adding poor weather conditions, and exploring additional sensors, such as Light Detection and Ranging (LiDAR), to enhance detection accuracy in complete darkness.