Testing Whether GANs Can Generate Photorealistic Images from Rough Sketches


by : 
Tianyue Z.


Generative adversarial networks (GANs) can generate images from rough sketches, potentially making image composition much more efficient than it is currently. However, the images from GANs are generally not photorealistic enough to be used in many real life applications. This project will attempt to make a GAN generate truly photorealistic images that can be used in real life applications.