Text to Realistic Image Generation with Attentional Concatenation Generative Adversarial Networks

Figure 5

Images generated from descriptions using three GAN models trained on Oxford test set.