A Review On Generating Synthetic Images From Textual Descriptions Using Semantic-Spatial Aware Gans

Abstract

he subject of Generative Adversarial Networks (GANs) has experienced tremendous advancements over the last ten years, especially in the creation of artificial images from textual descriptions. Due to its capacity to comprehend and apply both semantic and spatial information in the text-to-image generation process, Semantic-Spatial Aware GANs have become one of the more potent methods among those in development. With a thorough comparison of the main approaches and their performance measures, this study examines the developments in Semantic-Spatial Aware GANs during the past decade.