Description of Invention
Text-to-Image synthesis will convert the given text and generate images related to the meaning extracted from the sentence. Text to image synthesis approaches mainly develop based on generative adversarial network (GAN) and some extensive methods from it have been published. Text-to-Image synthesis could be used in diverse applications including real-time image generation, image manipulation, and image illustration based on text descriptions. In this research project, self-supervised learning is used to explore the high-level structural information in synthesising complex objects.