tf_models.py

import tensorflow as tf
from tensorflow.keras import layers

def downsample(filters, size, apply_instancenorm=True):
    initializer = tf.random_normal_initializer(0., 0.02)
    result = tf.keras.Sequential()
    result.add(layers.Conv2D(filters, size, strides=2, padding='same',
                             kernel_initializer=initializer, use_bias=False))
    if apply_instancenorm:
        result.add(tf.keras.layers.GroupNormalization(groups=-1))
    result.add(layers.LeakyReLU())
    return result

def upsample(filters, size, apply_dropout=False):
    initializer = tf.random_normal_initializer(0., 0.02)
    result = tf.keras.Sequential()
    result.add(layers.Conv2DTranspose(filters, size, strides=2, padding='same',
                                      kernel_initializer=initializer, use_bias=False))
    result.add(tf.keras.layers.GroupNormalization(groups=-1))
    if apply_dropout:
        result.add(layers.Dropout(0.5))
    result.add(layers.ReLU())
    return result

def resnet_block(filters, size=3):
    initializer = tf.random_normal_initializer(0., 0.02)
    result = tf.keras.Sequential()
    result.add(layers.Conv2D(filters, size, padding='same', kernel_initializer=initializer, use_bias=False))
    result.add(tf.keras.layers.GroupNormalization(groups=-1))
    result.add(layers.ReLU())
    result.add(layers.Conv2D(filters, size, padding='same', kernel_initializer=initializer, use_bias=False))
    result.add(tf.keras.layers.GroupNormalization(groups=-1))
    return result

def Generator(output_channels=3, num_resnet=9):
    inputs = layers.Input(shape=[256, 256, 3])
    
    # Downsampling
    x = layers.Conv2D(64, 7, padding='same', kernel_initializer=tf.random_normal_initializer(0., 0.02), use_bias=False)(inputs)
    x = tf.keras.layers.GroupNormalization(groups=-1)(x)
    x = layers.ReLU()(x)
    
    x = downsample(128, 3)(x)
    x = downsample(256, 3)(x)
    
    # Residual blocks
    for _ in range(num_resnet):
        res = resnet_block(256)(x)
        x = layers.Add()([x, res])
        
    # Upsampling
    x = upsample(128, 3)(x)
    x = upsample(64, 3)(x)
    
    last = layers.Conv2D(output_channels, 7, padding='same', activation='tanh', 
                          kernel_initializer=tf.random_normal_initializer(0., 0.02))(x)
    
    return tf.keras.Model(inputs=inputs, outputs=last)

def Discriminator():
    initializer = tf.random_normal_initializer(0., 0.02)
    inputs = layers.Input(shape=[256, 256, 3])
    
    down1 = downsample(64, 4, False)(inputs) # (bs, 128, 128, 64)
    down2 = downsample(128, 4)(down1)       # (bs, 64, 64, 128)
    down3 = downsample(256, 4)(down2)       # (bs, 32, 32, 256)
    
    zero_pad1 = layers.ZeroPadding2D()(down3) # (bs, 34, 34, 256)
    conv = layers.Conv2D(512, 4, strides=1, kernel_initializer=initializer, use_bias=False)(zero_pad1)
    norm1 = tf.keras.layers.GroupNormalization(groups=-1)(conv)
    leaky_relu = layers.LeakyReLU()(norm1)
    
    zero_pad2 = layers.ZeroPadding2D()(leaky_relu)
    last = layers.Conv2D(1, 4, strides=1, kernel_initializer=initializer)(zero_pad2)
    
    return tf.keras.Model(inputs=inputs, outputs=last)