Performance Analysis of Deep Learning RBFNN Algorithm Based Bridge/Concrete Detection

Identifying structural damage to commercial and industrial infrastructures (such as buildings, bridges, and roads) depends on being able to spot cracks, which are the first sign of serious concrete structural degradation. The classification of the fracture injury is the first and most important step. Unfortunately, because of the subjectivity element, this procedure is time-consuming and may result in uncertainty. Deep learning and image processing advancements may now allow a machine to identify a bridge/concrete problem automatically. The Concrete image will be pre-processed using the Adaptive Median Filter (AMF), which is intended to enhance photographs. Since Fuzzy C-Means (FCM) allows the clustering process to retain more original image information than crisp or hard clustering approaches, it has been recognized as a reliable picture segmentation algorithm. Grey Level Co-occurrence Matrix (GLCM) is used to extract second order texture information from photos and Radial Basis Function Neural Network (RBFNN) is used as an object classification to construct a machine learning-based road fracture detection system. After 20 epochs, the Training Loss is 0.88 and the Training Validation Loss is 5.