Chapter 6: Deep learning-based surface crack detection in fibre-reinforced concrete exposed to temperature variations

Fibre-reinforced concrete (FRC) is a ductile material used for structural applications, especially in bending members. The application of FRC has become vital in the construction sector. Surface cracks in FRC pose significant challenges to structural integrity, particularly when subjected to temperature fluctuations. Conventional techniques for detecting cracks depend on visual inspection, which is labourious and prone to human mistakes. This chapter explores the utilisation of deep learning methods to identify surface cracks in fibre-reinforced concrete under varying temperature conditions. Steel fibre, basalt fibre and polypropylene fibre were used to prepare the specimens. The concrete samples were applied under different ranges of heating durations. Compressive strength, mass loss and temperature variation at the core of the specimens were evaluated and reported. A faster region-based convolutional neural network (RCNN) technique was used to find the crack regions. The quantitative measure of cracks was obtained through distance vector transform. The goal of the research presented in this chapter was to create a reliable and effective system capable of precisely detecting and characterising surface cracks, facilitating prompt maintenance and repair actions. The proposed technique was validated with numerical correlation analysis. From the experimental results, it is understood that faster RCNN detects concrete surface cracks efficiently.