This repository contains the code for the CS726 Project
The objective of the code is to generate images from a text description using a Generative Adversarial Network.
This implementation closely follows the Stage I implementation from the paper https://arxiv.org/abs/1612.03242
The major changes from the conventional conditional GAN and previous Text to Image synthesis paper is the use of a Conditional Augmentation technique adapted from the VAE paper which is based on Reparametrization Trick.