Devising Ceramic Tiles Texture featuring Semi-automatic Image Clustering and Generative Adversarial Networks

Abstract

Neural network-based generation has become a promising area of research with diverse applications in recent years, such as generating synthetic data, medical images, celebrity faces and so on. In this study, we try to investigate the potential of generative models in the field of ceramic tiles design. Specifically, we employ DCGAN and WGAN-GP to generate textures of ceramic tiles. Our dataset undergoes a semi-automated pre-processing stage, which has been crucial for achieving high-quality outputs. We analyse the performances and outputs of both the models. To examine the spatial quality of the produced outputs, employing statistical calculation and collecting survey reviews from professionals in the ceramic tile industry have been done to gather better insights. Our results demonstrate the potential of generative models for producing realistic ceramic tiles textures and suggest this work holds promise for improving the creativity and efficiency of ceramic tiles design process.