In manufacturing industry, specially (ceramic tiles industry) manpower are doing manual inspection for detecting the defects. This process is very tedious and expensive, while detecting the flaws manually; there is a risk of getting inaccuracy in this work. The number of the tiles produced in real time environment, it is impossible to detect the defects in manual mode. The Ceramic tiles industry facing a lot of challenges in defect detection due to shape of defect, location of the defect, environment conditions and lighting conditions. Deep learning is a sub set of machine learning which helps to take the decision based on the tiles. We have proposed four deep learning approaches (Pre-trained models) namely, Dense Net, Xception, Res Net50 and VGG16 in this work. In this study, all four techniques have been thoroughly explored. Consequently, all four models are capable of learning a wide range of images and identifying flaws in ceramic tiles.