Deep Learning-Based Damage Inspection for Concrete Structures

Abstract

It is important to detect and assess the damaged condition of concrete structures to ensure their safety. Several common indicators of the damaged condition such as dimension deviation of precast concrete (PC) components, propagation of the local crack, anomaly of vibration response, and decrease in structural bearing capacity are difficult to be detected automatically and efficiently. The rapid development of deep learning (DL) provides new ideas to solve the above problems. DL can effectively extract features from a large amount of multi-source data such as images, point clouds, and acceleration data, and perform related classification and detection tasks. The authors demonstrate the potential of DL in engineering practice by employing DL to address the crucial problems of damage inspection for concrete structures.