Abstract

Dams are critical components of national infrastructure, and the presence of undetected surface cracks can lead to severe safety hazards. Conventional inspection methods are often time-consuming, limited by human error, and difficult to implement on a large scale, making timely assessments a persistent challenge. To overcome these limitations, this study presents an automated crack detection framework that leverages YOLOv10 in conjunction with UAV-captured imagery. The speed and accuracy of YOLOv10, combined with the UAV’s ability to access hard-to-reach areas, significantly enhance the efficiency and reliability of the inspection process. The proposed system achieved promising results, recording a precision of 0.93, a recall of 0.81, and an F1-score of 0.86— reflecting its capability to rapidly and accurately detect surface-level cracks. This approach represents a step forward in advancing intelligent infrastructure management by facilitating early maintenance and contributing to safer and more sustainable dam operations, particularly through its potential integration with digital twin technologies.