Deep Learning for Computer Vision

Conditional GAN

A conditional GAN generates images with a label that we want. For example, we can ask the model to produce the digit 8 and the model will produce an 8. To enable this, the labels are required along with noise to be trained with the model, as shown here:

gan = tf.contrib.gan.gan_model(    get_generator,    get_discriminator,    real_images,    (input_noise, labels))

The rest of the training is similar to that of the vanilla GAN. Next, we will use a GAN for compressing the images.