Surface Crack Detection Using Deep Learning Models

Surface cracks are a widespread issue that affects many different businesses, including the building industry, the transportation industry, and the manufacturing industry. In order to preserve safety, minimize expensive repairs, and ensure the lifespan of buildings and equipment, it is essential to detect and identify surface cracks. The research made use of a dataset of surface cracks that is freely accessible to the public from Kaggle website. A level of accuracy of 99.62% was attained by the VGG16 model. The accuracy of the Inception model was measured at 99.89%. The Xception model was successful in achieving an accuracy of 99.76%. According to these findings, the Inception model performed far better than the other two models when it came to identifying surface cracks.