Generative adversarial networks: A recent survey

Abstract

The enormous potential of GANs has enabled their usage for a variety of diverse applications. In this survey, we highlight major applications that have been made throughout the years, including the creation of images for the visualization of stories, the repair of damaged or old photographs, the creation of 3D models, the translation of text into images, and more. The survey indicates several constraints and barriers to more consistent training as well as creating higher-quality pictures. For instance, some models may struggle to eliminate deformities like wave strikes. As GAN development goes on, there will be more chances for advancement and the emergence of new applications.