Evaluating YOLOv8 for road damage detection

This research investigates the use of YOLOv8 model, renowned for its speed and accuracy to develop an automated road damage detection system. Roads are a critical part of transportation infrastructure, contributing significantly to social and economic development, yet they are prone to damages which can lead to considerable financial and safety concerns. Traditional manual inspections are often inefficient and error prone, creating a need for more advanced and reliable detection methods. Using the Road Damage Dataset (RDD2022), the YOLOv8 model was trained to identify multiple types of road damage. Evaluation using standard metrics demonstrates the model’s effectiveness, though areas for improvement remain. The results highlight YOLOv8’s potential to enhance road maintenance operations and improve road safety.