Generative Adversarial Network (GAN) to Generate Realistic Images

Abstract

Abstract: Generative Adversarial Networks (GANs) have rapidly become a focal point of research due to their ability to generate realistic images. First introduced in 2014, GANs have been applied in a multitude of fields such as computer vision and natural language processing, yielding impressive results. Image synthesis is among the most thoroughly researched applications of GANs, and the results thus far have demonstrated the potential of GANs in image synthesis. This paper provides a taxonomy of the methods used in image synthesis, reviews various models for text-to-image synthesis and image-to-image translation, discusses evaluation metrics, and highlights future research directions for image synthesis using GANs..