Enhancing Structural Health Monitoring: AI-Driven Image Processing for Automated Crack Identification in Concrete Surfaces

One of the first signs of structural deterioration is cracks in the concrete surface, which is important for maintenance because prolonged exposure will seriously harm the environment. The highly regarded method for inspecting cracks is manual inspection. During the handwritten inspection, the irregularities' conditions are noted and a hand-drawn illustrate of the crack is created. One of the main problems on large construction sites is the degradation of structures as a result of cracks, especially in cases where inspection by hand is not feasible. Thus, an alternative called automatic image-based crack detection is suggested. The primary goal of this research work is to use artificial intelligence-based image processing techniques to identify and categorize cracks. Here, this study proposes five modules. The first module collects the dataset of crack images; the second module applies a pre-processing method to filter the data; the third module divides the data into training and testing portions; and the fourth module implements a model that uses both deep learning, such as the Mobile-net algorithm, and machine learning, such as the Random Forest algorithm, for testing and training the data. The last part is the prediction module, which forecasts whether a crack image will be positive or negative. Lastly, this study compares the accuracy score, recall, precision, loss graph, F1 score value, and algorithms for machine learning along with deep learning. The system performed better, as demonstrated by the experimental results.