Quiz: Image Restoration with GANs

This chapter helped us in performing image super-resolution with SRGAN to generate high-resolution images from low-resolution ones and use a data prefetcher to speed up data loading and increase our GPU’s efficiency during training. We also learned how to implement our own convolution with several methods, including the direct approach, the FFT-based method, and the im2col method.

We also got to see the disadvantages of vanilla GAN loss functions and how to improve them by using Wasserstein loss (the Wasserstein GAN). By the end of this section, we had learned how to train a GAN model to perform image inpainting and fill in the missing parts of an image, too.

Test your understanding

You can attempt the quiz below to test your understanding: