Text to Realistic Image Generation with Attentional Concatenation Generative Adversarial Networks

Figure 4

Details (beak, wings) comparison of the images generated from descriptions using three GAN models trained on CUB test set.