Generative Adversarial Networks Projects

Training the discriminator networks

The code inside this sub-section is a continuation of the code in the last section. Here, you will see how to train the discriminator networks:

1. Start by sampling a mini-batch of images for both domains, as shown in the following code:

        batchA = imagesA[index * batch_size:(index + 1) * batch_size]        batchB = imagesB[index * batch_size:(index + 1) * batch_size]

2. Next, generate fake images using the generator networks, as follows:

        generatedB = generatorAToB.predict(batchA)        generatedA = generatorBToA.predict(batchB)

3. After that, train the discriminator network A on both real and fake images (generated by the generator network), as follows:

        dALoss1 = discriminatorA.train_on_batch(batchA, real_labels)