還沒有女朋友的朋友們，你們有福了，學會CycleGAN把男朋友變成女朋友

前言

事情的起因是這樣的，室友在經歷的4年的找女朋友之旅后，終于放棄了，而我為了讓他的青春不留遺憾，只能使用 CycleGAN 把下鋪壯漢變成萌妹了，
轉眼又到了畢業季，還在為沒有女朋友而著急么？還在為沒有談一場青春的戀愛而遺憾么？還沒有女朋友的朋友們，你們有福了！！！沒有女朋友，還能沒有男朋友么？學會 CycleGAN ，把男朋友變成女朋友，趕快學起來吧，

效果展示

在學習之前，大家肯定想先知道CycleGAN模型進行男女性別轉換的效果如何，所以先讓大家看看模型訓練的效果.

效果這么驚人，還不快學起來？？？

使用 CycleGAN 進行不成對的影像轉換

CycleGAN 可以使用兩個生成器和兩個鑒別器訓練不成對(unpaired)的影像，
本文主要以實戰為主，如果想要了解 CycleGAN 背后的具體原理，請參考 CycleGAN 原理與實作（采用tensorflow2.x實作）.

不成對的資料集

CycleGAN 的一個重要貢獻是，改變了pix2pix需要成對的訓練資料集的缺點，某些情況下，我們可以很容易地創建成對資料集，如彩色的影像對應的灰度影像資料集，完成成對資料集的構建，從而用于訓練灰度影像上色的深度學習模型，但是，更多數的情況下，無法創建成對的資料集，例如從男性到女性的影像轉換，
這便是 CycleGAN 的優勢所在，因為它不需要成對的資料， CycleGAN 可以訓練不成對的資料集！

CycleGAN模型

簡單看下CycleGAN的體系架構：

資料集

資料集取自 Celeb A ，可以自行構建資料集，也可以使用此資料集，提取碼：nql9，

資料加載與預處理

# 匯入必要庫
import tensorflow as tf
import os
import time
from matplotlib import pyplot as plt
import tensorflow_datasets as tfds
AUTOTUNE = tf.data.experimental.AUTOTUNE

# 定義超引數
BUFFER_SIZE = 128
BATCH_SIZE = 1
IMG_WIDTH = 256
IMG_HEIGHT = 256
OUTPUT_CHANNELS = 3
LAMBDA = 10
EPOCHS = 100
"""
# 資料預處理函式
"""
def random_crop(image):
    cropped_image = tf.image.random_crop(image, size=[IMG_HEIGHT, IMG_WIDTH, 3])
    return cropped_image

# normalizing the images to [-1, 1]
def normalize(image):
    image = tf.cast(image, tf.float32)
    image = (image / 127.5) - 1
    return image

def random_jitter(image):
    # resizing to 286 x 286 x 3
    image = tf.image.resize(image, [286, 286], method=tf.image.ResizeMethod.NEAREST_NEIGHBOR)
    # randomly cropping to 256 x 256 x 3
    image = random_crop(image)
    # random mirroring
    image = tf.image.random_flip_left_right(image)
    return image

def preprocess_image_train(image):
    image = random_jitter(image)
    image = normalize(image)
    return image

def preprocess_image_test(image):
    image = normalize(image)
    return image

def load(image_file):
    image = tf.io.read_file(image_file)
    image = tf.image.decode_jpeg(image)
    input_image = tf.cast(image, tf.float32)
    return input_image

def load_image_train(image_file):
    image = load(image_file)
    image = preprocess_image_train(image)
    return image

def load_image_test(image_file):
    image = load(image_file)
    image = preprocess_image_test(image)
    return image

# 加載男性圖片，構建訓練資料集
train_man = tf.data.Dataset.list_files('./man2woman/trainA/*.jpg')
train_man = train_man.map(load_image_train, num_parallel_calls=tf.data.experimental.AUTOTUNE)
train_man = train_man.shuffle(BUFFER_SIZE)
train_man = train_man.batch(BATCH_SIZE, drop_remainder=True)
# 加載女性圖片，構建訓練資料集
train_woman = tf.data.Dataset.list_files('./man2woman/trainB/*.jpg')
train_woman = train_woman.map(load_image_train, num_parallel_calls=tf.data.experimental.AUTOTUNE)
train_woman = train_woman.shuffle(BUFFER_SIZE)
train_woman = train_woman.batch(BATCH_SIZE, drop_remainder=True)

模型構建

在 CycleGAN 中，使用實體歸一化代替批歸一化，但在 tensorflow 中，未包含實體歸一化層，因此需要自行實作，

class InstanceNormalization(tf.keras.layers.Layer):
    """Instance Normalization Layer."""

    def __init__(self, epsilon=1e-5):
        super(InstanceNormalization, self).__init__()
        self.epsilon = epsilon

    def build(self, input_shape):
        self.scale = self.add_weight(
            name='scale', 
            shape=input_shape[-1:],
            initializer=tf.random_normal_initializer(1., 0.02),
            trainable=True)
        self.offset = self.add_weight(
            name='offset',
            shape=input_shape[-1:],
            initializer='zeros',
            trainable=True)

    def call(self, x):
        mean, variance = tf.nn.moments(x, axes=[1, 2], keepdims=True)
        inv = tf.math.rsqrt(variance + self.epsilon)
        normalized = (x - mean) * inv
        return self.scale * normalized + self.offset

為了減少代碼量，定義上采樣塊和下采樣塊：

# 下采樣塊
def downsample(filters, size, norm_type='batchnorm', apply_norm=True):
    initializer = tf.random_normal_initializer(0., 0.02)

    result = tf.keras.Sequential()
    result.add(
        tf.keras.layers.Conv2D(filters, size, strides=2, padding='same',
                                kernel_initializer=initializer, use_bias=False))
    if apply_norm:
        if norm_type.lower() == 'batchnorm':
            result.add(tf.keras.layers.BatchNormalization())
        elif norm_type.lower() == 'instancenorm':
            result.add(InstanceNormalization())
    result.add(tf.keras.layers.LeakyReLU())
    return result

# 上采樣快
def upsample(filters, size, norm_type='batchnorm', apply_dropout=False):
    initializer = tf.random_normal_initializer(0., 0.02)
    result = tf.keras.Sequential()
    result.add(
        tf.keras.layers.Conv2DTranspose(filters, size, strides=2,
                                        padding='same',
                                        kernel_initializer=initializer,
                                        use_bias=False))
    if norm_type.lower() == 'batchnorm':
        result.add(tf.keras.layers.BatchNormalization())
    elif norm_type.lower() == 'instancenorm':
        result.add(InstanceNormalization())
    if apply_dropout:
        result.add(tf.keras.layers.Dropout(0.5))
    result.add(tf.keras.layers.ReLU())
    return result

接下來構建生成器：

def unet_generator(output_channels, norm_type='batchnorm'):
    down_stack = [
        downsample(64, 4, norm_type, apply_norm=False), 
        downsample(128, 4, norm_type),
        downsample(256, 4, norm_type),
        downsample(512, 4, norm_type),
        downsample(512, 4, norm_type),
        downsample(512, 4, norm_type),
        downsample(512, 4, norm_type),
        downsample(512, 4, norm_type),
    ]

    up_stack = [
        upsample(512, 4, norm_type, apply_dropout=True),
        upsample(512, 4, norm_type, apply_dropout=True),
        upsample(512, 4, norm_type, apply_dropout=True),
        upsample(512, 4, norm_type),
        upsample(256, 4, norm_type),
        upsample(128, 4, norm_type),
        upsample(64, 4, norm_type),
    ]

    initializer = tf.random_normal_initializer(0., 0.02)
    last = tf.keras.layers.Conv2DTranspose(
        output_channels, 4, strides=2,
        padding='same', kernel_initializer=initializer,
        activation='tanh')  # (bs, 256, 256, 3)
    concat = tf.keras.layers.Concatenate()
    inputs = tf.keras.layers.Input(shape=[None, None, 3])
    x = inputs
    # Downsampling through the model
    skips = []
    for down in down_stack:
        x = down(x)
        skips.append(x)
    skips = reversed(skips[:-1])
    # Upsampling and establishing the skip connections
    for up, skip in zip(up_stack, skips):
        x = up(x)
        x = concat([x, skip])
    x = last(x)
    return tf.keras.Model(inputs=inputs, outputs=x)

構建鑒別器：

def discriminator(norm_type='batchnorm', target=True):
    initializer = tf.random_normal_initializer(0., 0.02)

    inp = tf.keras.layers.Input(shape=[None, None, 3], name='input_image')
    x = inp

    if target:
        tar = tf.keras.layers.Input(shape=[None, None, 3], name='target_image')
        x = tf.keras.layers.concatenate([inp, tar])  # (bs, 256, 256, channels*2)

    down1 = downsample(64, 4, norm_type, False)(x)  # (bs, 128, 128, 64)
    down2 = downsample(128, 4, norm_type)(down1)  # (bs, 64, 64, 128)
    down3 = downsample(256, 4, norm_type)(down2)  # (bs, 32, 32, 256)

    zero_pad1 = tf.keras.layers.ZeroPadding2D()(down3)  # (bs, 34, 34, 256)
    conv = tf.keras.layers.Conv2D(
        512, 4, strides=1, kernel_initializer=initializer,
        use_bias=False)(zero_pad1)  # (bs, 31, 31, 512)
    if norm_type.lower() == 'batchnorm':
        norm1 = tf.keras.layers.BatchNormalization()(conv)
    elif norm_type.lower() == 'instancenorm':
        norm1 = InstanceNormalization()(conv)
    leaky_relu = tf.keras.layers.LeakyReLU()(norm1)
    zero_pad2 = tf.keras.layers.ZeroPadding2D()(leaky_relu)  # (bs, 33, 33, 512)
    last = tf.keras.layers.Conv2D(
        1, 4, strides=1,
        kernel_initializer=initializer)(zero_pad2)  # (bs, 30, 30, 1)

    if target:
        return tf.keras.Model(inputs=[inp, tar], outputs=last)
    else:
        return tf.keras.Model(inputs=inp, outputs=last)

實體化生成器與鑒別器：

generator_g = unet_generator(OUTPUT_CHANNELS, norm_type='instancenorm')
generator_f = unet_generator(OUTPUT_CHANNELS, norm_type='instancenorm')
discriminator_x = discriminator(norm_type='instancenorm', target=False)
discriminator_y = discriminator(norm_type='instancenorm', target=False)

損失函式與優化器的定義：

loss_obj = tf.keras.losses.BinaryCrossentropy(from_logits=True)
# 鑒別器損失
def discriminator_loss(real, generated):
    real_loss = loss_obj(tf.ones_like(real), real)
    generated_loss = loss_obj(tf.zeros_like(generated), generated)
    total_disc_loss = real_loss + generated_loss
    return total_disc_loss * 0.5
# 生成器損失
def generator_loss(generated):
    return loss_obj(tf.ones_like(generated), generated)
# 回圈一致性損失
def calc_cycle_loss(real_image, cycled_image):
    loss1 = tf.reduce_mean(tf.abs(real_image - cycled_image))
    return LAMBDA * loss1
# identity loss
def identity_loss(real_image, same_image):
    loss = tf.reduce_mean(tf.abs(real_image - same_image))
    return LAMBDA * 0.5 * loss
# 優化器
generator_g_optimizer = tf.keras.optimizers.Adam(2e-4, beta_1=0.5)
generator_f_optimizer = tf.keras.optimizers.Adam(2e-4, beta_1=0.5)
discriminator_x_optimizer = tf.keras.optimizers.Adam(2e-4, beta_1=0.5)
discriminator_y_optimizer = tf.keras.optimizers.Adam(2e-4, beta_1=0.5)

訓練結果可視化函式

創建 generate_images 函式用于在訓練程序中查看模型效果.

def generate_images(model, test_input):
    prediction = model(test_input)

    plt.figure(figsize=(12, 12))

    display_list = [test_input[0], prediction[0]]
    title = ['Input Image', 'Predicted Image']

    for i in range(2):
        plt.subplot(1, 2, i+1)
        plt.title(title[i])
        # getting the pixel values between [0, 1] to plot it.
        plt.imshow(display_list[i] * 0.5 + 0.5)
        plt.axis('off')
    # plt.show()
    plt.savefig('results/{}.png'.format(int(time.time())))

訓練步驟

首先需要定義訓練函式：

@tf.function
def train_step(real_x, real_y):
    with tf.GradientTape(persistent=True) as tape:
        # Generator G translates X -> Y
        # Generator F translates Y -> X.
        fake_y = generator_g(real_x, training=True)
        cycled_x = generator_f(fake_y, training=True)
        fake_x = generator_f(real_y, training=True)
        cycled_y = generator_g(fake_x, training=True)

        # same_x and same_y are used for identity loss.
        same_x = generator_f(real_x, training=True)
        same_y = generator_g(real_y, training=True)

        disc_real_x = discriminator_x(real_x, training=True)
        disc_real_y = discriminator_y(real_y, training=True)

        disc_fake_x = discriminator_x(fake_x, training=True)
        disc_fake_y = discriminator_y(fake_y, training=True)

        # calculate the loss
        gen_g_loss = generator_loss(disc_fake_y)
        gen_f_loss = generator_loss(disc_fake_x)

        total_cycle_loss = calc_cycle_loss(real_x, cycled_x) + calc_cycle_loss(real_y, cycled_y)

        # Total generator loss = adversarial loss + cycle loss
        total_gen_g_loss = gen_g_loss + total_cycle_loss + identity_loss(real_y, same_y)
        total_gen_f_loss = gen_f_loss + total_cycle_loss + identity_loss(real_x, same_x)

        disc_x_loss = discriminator_loss(disc_real_x, disc_fake_x)
        disc_y_loss = discriminator_loss(disc_real_y, disc_fake_y)

    # Calculate the gradients for generator and discriminator
    generator_g_gradients = tape.gradient(total_gen_g_loss, generator_g.trainable_variables)
    generator_f_gradients = tape.gradient(total_gen_f_loss, generator_f.trainable_variables)
    discriminator_x_gradients = tape.gradient(disc_x_loss, discriminator_x.trainable_variables)
    discriminator_y_gradients = tape.gradient(disc_y_loss, discriminator_y.trainable_variables)

    # Apply the gradients to the optimizer
    generator_g_optimizer.apply_gradients(zip(generator_g_gradients, generator_g.trainable_variables))
    generator_f_optimizer.apply_gradients(zip(generator_f_gradients, generator_f.trainable_variables))
    discriminator_x_optimizer.apply_gradients(zip(discriminator_x_gradients, discriminator_x.trainable_variables))
    discriminator_y_optimizer.apply_gradients(zip(discriminator_y_gradients, discriminator_y.trainable_variables))

最后進行模型的訓練：

for epoch in range(EPOCHS):
    start = time.time()
    
    n = 0
    for image_x, image_y in tf.data.Dataset.zip((train_man, train_woman)):
        train_step(image_x, image_y)
        # generate_images(generator_g, sample_man)
        if n % 10 == 0:
            print ('.', end='')
            n += 1
	# 采樣測驗資料集， 測驗模型效果
	sample_man = next(iter(train_man))
	sample_woman = next(iter(train_woman))
    generate_images(generator_g, sample_man)
    generate_images(generator_f, sample_woman)

    print ('Time taken for epoch {} is {} sec\n'.format(epoch + 1, time.time()-start))

效果二次展示

我們已經在開始時看到了 CycleGAN 在男性轉換為女性的結果，再看下將女性轉換為男性的效果吧！