Ceramic tiles are in high demand in the infrastructure and building development industries due to their low cost, ease of installation, maintenance, moisture resistance, and availability in a broad range of colors, textures, and sizes. Automated facilities, which produce hundreds of tiles in every segment, require a tremendous volume of output. Because of the large number of tiles produced and the frequency with which they are produced, it is impossible to manually examine them for faults, necessitating the use of a rapid, efficient, and reliable automated process. However, while the process of detecting flaws and categorizing them (or classification) is not as efficient as it might be, recent advances in computing technology, mathematical modeling, and high-resolution picture capture equipment have given rise to new prospects in the subject. Many kinds of literature on using these systems for the same goal are currently accessible. Deep learning is a type of artificial intelligence that helps people makes decisions. In production applications, image detection of faulty Ceramic Tile Surfaces is a critical skill. Deep learning is now being studied for its potential application in automated defect identification. As a result, we propose Deep Learning approaches that take advantage of the transform domain properties of the tiles image. The model's capacity to learn via the system makes it versatile and dynamically customizable. Different deep learning-based fault detection and classification transfer learning approaches are examined in this study

Keywords: Deep learning, Classification, Ceramic tile images, VGG model, CNN