There is a global research trend to enhance condition assessment of the concrete infrastructure by the development of advanced nondestructive testing (NDT) methods. Computer vision–based systems have been developed to detect different types of defects in both regular and thermographic images because these systems could offer a timely and cost-effective solution and are able to tackle the inconsistency issues of manual assessment. This paper investigates the performance of different deep neural network models to detect main concrete anomalies, including delamination, cracks, spalling, and patches in thermographic and regular images captured from a variety of distances and viewpoints. These models were trained and tested using images taken from a century-old buttress dam and validated in images captured from the decks of two concrete bridges. The results showed that the MobileNetV2 had promising performance in the identification of multiclass damages in the thermal images, identifying 79.7% of the total delamination, cracks, spalling, and patches on the test images of highly damaged concrete areas. The VGG 16 model showed better precision by reducing the number of false detections.

One of the concerns about the use of passive Infrared Thermography (IRT) for structural health monitoring (SHM) is the determination of a favorable period to conduct the inspections. This paper investigates the use of numerical simulations to find appropriate periods for IRT-based detection of subsurface damages in concrete bridge slabs under passive heating along a 1 year of time span. A model was built using the Finite Element Method (FEM) and calibrated using the results of a set of thermographic field inspections on a concrete slab sample. The results showed that the numerical simulation properly reproduced the experimental thermographic measurements of the concrete structure under passive heating, allowing the analysis to be extended for a longer testing period. The long-term FEM results demonstrated that the months of spring and summer are the most suitable for passive IRT inspections in this study, with around 17% more detections compared to the autumn and winter periods in Brazil. By enhancing the possibility of using FEM beyond the design stage, we demonstrate that this computation tool can provide support to long-term SHM.