cyclegan_discriminator

cyclegan_discriminator.py

def make_discriminator_model():
    # C64-C128-C256-C512
    model = tf.keras.Sequential()
    model.add(tf.keras.layers.Conv2D(64, (4, 4), strides=(2, 2), padding='same', input_shape=(256, 256, 3)))
    model.add(tf.keras.layers.LeakyReLU(alpha=0.2))

    model.add(tf.keras.layers.Conv2D(128, (4, 4), strides=(2, 2), padding='same', use_bias=False))
    model.add(tf.keras.layers.BatchNormalization())
    model.add(tf.keras.layers.LeakyReLU(alpha=0.2))

    model.add(tf.keras.layers.Conv2D(256, (4, 4), strides=(2, 2), padding='same', use_bias=False))
    model.add(tf.keras.layers.BatchNormalization())
    model.add(tf.keras.layers.LeakyReLU(alpha=0.2))

    model.add(tf.keras.layers.Conv2D(512, (4, 4), strides=(1, 1), padding='same', use_bias=False))
    model.add(tf.keras.layers.BatchNormalization())
    model.add(tf.keras.layers.LeakyReLU(alpha=0.2))

    # This last conv net is the PatchGAN
    # https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix/issues/39#issuecomment-305575964
    # https://github.com/phillipi/pix2pix/blob/master/scripts/receptive_field_sizes.m
    model.add(tf.keras.layers.Conv2D(1, (4, 4), strides=(1, 1), padding='same'))

    return model