“…There is a large variety of computational methods that can be applied, such as interpolation functions, a histogram of oriented gradients (HOG), a feature descriptorsbased multi-layer perceptron (MLP) classifier, a support vector machine (SVM), a convolutional neural network (CNN), the limited and simplified hyperbolic summation (LSHS) technique, partial differential equations (PDEs), the hidden Markov model (HMM), a genetic algorithm, and the fuzzy logic approach. These computational methods are described and applied in some relevant and recent studies [177][178][179][180][181] and, in most cases, with accuracies greater than 95.0%. On the other hand, algorithms are also implemented to extract the GPR amplitudes automatically from both the whole GPR signals and the rebar peaks detected [165], [171], [174], generally after applying a migration procedure.…”