CycleGAN

import numpy as np
import matplotlib.pyplot as plt

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers

import tensorflow_addons as tfa
import tensorflow_datasets as tfds

tfds.disable_progress_bar()
autotune = tf.data.AUTOTUNE

# 使用tensorflow-datasets加载马-斑马数据集。
dataset, _ = tfds.load("cycle_gan/horse2zebra", with_info=True, as_supervised=True)
train_horses, train_zebras = dataset["trainA"], dataset["trainB"]
test_horses, test_zebras = dataset["testA"], dataset["testB"]

# 定义标准图像大小。
orig_img_size = (286, 286)
# 训练期间使用的随机裁剪大小。
input_img_size = (256, 256, 3)
# 层的权重初始化器。
kernel_init = keras.initializers.RandomNormal(mean=0.0, stddev=0.02)
# 实例归一化的Gamma初始化器。
gamma_init = keras.initializers.RandomNormal(mean=0.0, stddev=0.02)

buffer_size = 256
batch_size = 1


def normalize_img(img):
    img = tf.cast(img, dtype=tf.float32)
    # 将值映射到范围[-1, 1]
    return (img / 127.5) - 1.0


def preprocess_train_image(img, label):
    # 随机翻转
    img = tf.image.random_flip_left_right(img)
    # 首先调整为原始大小
    img = tf.image.resize(img, [*orig_img_size])
    # 随机裁剪到256X256
    img = tf.image.random_crop(img, size=[*input_img_size])
    # 将像素值归一化到范围[-1, 1]
    img = normalize_img(img)
    return img


def preprocess_test_image(img, label):
    # 测试图像只进行调整大小和归一化。
    img = tf.image.resize(img, [input_img_size[0], input_img_size[1]])
    img = normalize_img(img)
    return img

# 将预处理操作应用于训练数据
train_horses = (
    train_horses.map(preprocess_train_image, num_parallel_calls=autotune)
    .cache()
    .shuffle(buffer_size)
    .batch(batch_size)
)
train_zebras = (
    train_zebras.map(preprocess_train_image, num_parallel_calls=autotune)
    .cache()
    .shuffle(buffer_size)
    .batch(batch_size)
)

# 将预处理操作应用于测试数据
test_horses = (
    test_horses.map(preprocess_test_image, num_parallel_calls=autotune)
    .cache()
    .shuffle(buffer_size)
    .batch(batch_size)
)
test_zebras = (
    test_zebras.map(preprocess_test_image, num_parallel_calls=autotune)
    .cache()
    .shuffle(buffer_size)
    .batch(batch_size)
)

_, ax = plt.subplots(4, 2, figsize=(10, 15))
for i, samples in enumerate(zip(train_horses.take(4), train_zebras.take(4))):
    horse = (((samples[0][0] * 127.5) + 127.5).numpy()).astype(np.uint8)
    zebra = (((samples[1][0] * 127.5) + 127.5).numpy()).astype(np.uint8)
    ax[i, 0].imshow(horse)
    ax[i, 1].imshow(zebra)
plt.show()

class ReflectionPadding2D(layers.Layer):
    """实现反射填充作为一个层。

    参数:
        padding(tuple): 空间维度的填充量。

    返回:
        一个与输入张量相同类型的填充张量。
    """
    def __init__(self, padding=(1, 1), **kwargs):
        self.padding = tuple(padding)
        super().__init__(**kwargs)

    def call(self, input_tensor, mask=None):
        padding_width, padding_height = self.padding
        padding_tensor = [
            [0, 0],
            [padding_height, padding_height],
            [padding_width, padding_width],
            [0, 0],
        ]
        return tf.pad(input_tensor, padding_tensor, mode="REFLECT")



def residual_block(
    x,
    activation,
    kernel_initializer=kernel_init,
    kernel_size=(3, 3),
    strides=(1, 1),
    padding="valid",
    gamma_initializer=gamma_init,
    use_bias=False,
):
  dim = x.shape[-1]
    input_tensor = x

    x = ReflectionPadding2D()(input_tensor)
    x = layers.Conv2D(
        dim,
        kernel_size,
        strides=strides,
        kernel_initializer=kernel_initializer,
        padding=padding,
        use_bias=use_bias,
    )(x)
    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
    x = activation(x)

    x = ReflectionPadding2D()(x)
    x = layers.Conv2D(
        dim,
        kernel_size,
        strides=strides,
        kernel_initializer=kernel_initializer,
        padding=padding,
        use_bias=use_bias,
    )(x)
    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
    x = layers.add([input_tensor, x])
    return x



def downsample(
    x,
    filters,
    activation,
    kernel_initializer=kernel_init,
    kernel_size=(3, 3),
    strides=(2, 2),
    padding="same",
    gamma_initializer=gamma_init,
    use_bias=False,
):
    x = layers.Conv2D(
        filters,
        kernel_size,
        strides=strides,
        kernel_initializer=kernel_initializer,
        padding=padding,
        use_bias=use_bias,
    )(x)
    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
    if activation:
        x = activation(x)
    return x



def upsample(
    x,
    filters,
    activation,
    kernel_size=(3, 3),
    strides=(2, 2),
    padding="same",
    kernel_initializer=kernel_init,
    gamma_initializer=gamma_init,
    use_bias=False,
):
    x = layers.Conv2DTranspose(
        filters,
        kernel_size,
        strides=strides,
        padding=padding,
        kernel_initializer=kernel_initializer,
        use_bias=use_bias,
    )(x)
    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
    if activation:
        x = activation(x)
    return x

c7s1-64 ==> 卷积块，`relu`激活，过滤器大小为7
d128 ====|
         |-> 2个下采样块
d256 ====|
R256 ====|
R256     |
R256     |
R256     |
R256     |-> 9个残差块
R256     |
R256     |
R256     |
R256 ====|
u128 ====|
         |-> 2个上采样块
u64  ====|
c7s1-3 => 最后一个卷积块，`tanh`激活，过滤器大小为7。

def get_resnet_generator(
    filters=64,
    num_downsampling_blocks=2,
    num_residual_blocks=9,
    num_upsample_blocks=2,
    gamma_initializer=gamma_init,
    name=None,
):
    img_input = layers.Input(shape=input_img_size, name=name + "_img_input")
    x = ReflectionPadding2D(padding=(3, 3))(img_input)
    x = layers.Conv2D(filters, (7, 7), kernel_initializer=kernel_init, use_bias=False)(
        x
    )
    x = tfa.layers.InstanceNormalization(gamma_initializer=gamma_initializer)(x)
    x = layers.Activation("relu")(x)

    # 下采样
    for _ in range(num_downsampling_blocks):
        filters *= 2
        x = downsample(x, filters=filters, activation=layers.Activation("relu"))

    # 残差块
    for _ in range(num_residual_blocks):
        x = residual_block(x, activation=layers.Activation("relu"))

    # 上采样
    for _ in range(num_upsample_blocks):
        filters //= 2
        x = upsample(x, filters, activation=layers.Activation("relu"))

    # 最后一个块
    x = ReflectionPadding2D(padding=(3, 3))(x)
    x = layers.Conv2D(3, (7, 7), padding="valid")(x)
    x = layers.Activation("tanh")(x)

    model = keras.models.Model(img_input, x, name=name)
    return model

def get_discriminator(
    filters=64, kernel_initializer=kernel_init, num_downsampling=3, name=None
):
    img_input = layers.Input(shape=input_img_size, name=name + "_img_input")
    x = layers.Conv2D(
        filters,
        (4, 4),
        strides=(2, 2),
        padding="same",
        kernel_initializer=kernel_initializer,
    )(img_input)
    x = layers.LeakyReLU(0.2)(x)

    num_filters = filters
    for num_downsample_block in range(3):
        num_filters *= 2
        if num_downsample_block < 2:
            x = downsample(
                x,
                filters=num_filters,
                activation=layers.LeakyReLU(0.2),
                kernel_size=(4, 4),
                strides=(2, 2),
            )
        else:
            x = downsample(
                x,
                filters=num_filters,
                activation=layers.LeakyReLU(0.2),
                kernel_size=(4, 4),
                strides=(1, 1),
            )

    x = layers.Conv2D(
        1, (4, 4), strides=(1, 1), padding="same", kernel_initializer=kernel_initializer
    )(x)

    model = keras.models.Model(inputs=img_input, outputs=x, name=name)
    return model


# 获取生成器
gen_G = get_resnet_generator(name="generator_G")
gen_F = get_resnet_generator(name="generator_F")

# 获取判别器
disc_X = get_discriminator(name="discriminator_X")
disc_Y = get_discriminator(name="discriminator_Y")

class CycleGan(keras.Model):
    def __init__(
        self,
        generator_G,
        generator_F,
        discriminator_X,
        discriminator_Y,
        lambda_cycle=10.0,
        lambda_identity=0.5,
    ):
        super().__init__()
        self.gen_G = generator_G
        self.gen_F = generator_F
        self.disc_X = discriminator_X
        self.disc_Y = discriminator_Y
        self.lambda_cycle = lambda_cycle
        self.lambda_identity = lambda_identity

    def compile(
        self,
        gen_G_optimizer,
        gen_F_optimizer,
        disc_X_optimizer,
        disc_Y_optimizer,
        gen_loss_fn,
        disc_loss_fn,
    ):
        super().compile()
        self.gen_G_optimizer = gen_G_optimizer
        self.gen_F_optimizer = gen_F_optimizer
        self.disc_X_optimizer = disc_X_optimizer
        self.disc_Y_optimizer = disc_Y_optimizer
        self.generator_loss_fn = gen_loss_fn
        self.discriminator_loss_fn = disc_loss_fn
        self.cycle_loss_fn = keras.losses.MeanAbsoluteError()
        self.identity_loss_fn = keras.losses.MeanAbsoluteError()

    def train_step(self, batch_data):
        # x 是马，y 是斑马
        real_x, real_y = batch_data

        # 对于 CycleGAN，我们需要为生成器和判别器计算不同的
        # 损失类型。我们将在这里执行以下步骤：
        #
        # 1. 将真实图像通过生成器，获取生成的图像
        # 2. 将生成的图像再传回生成器，以检查我们是否
        #    可以从生成的图像预测出原始图像。
        # 3. 使用生成器对真实图像进行身份映射。
        # 4. 将步骤 1) 中生成的图像传递给相应的判别器。
        # 5. 计算生成器的总损失（对抗损失 + 循环损失 + 身份损失）
        # 6. 计算判别器的损失
        # 7. 更新生成器的权重
        # 8. 更新判别器的权重
        # 9. 在字典中返回损失

        with tf.GradientTape(persistent=True) as tape:
            # 马到假斑马
            fake_y = self.gen_G(real_x, training=True)
            # 斑马到假马 -> y2x
            fake_x = self.gen_F(real_y, training=True)

            # 循环（马到假斑马再到假马）：x -> y -> x
            cycled_x = self.gen_F(fake_y, training=True)
            # 循环（斑马到假马再到假斑马）：y -> x -> y
            cycled_y = self.gen_G(fake_x, training=True)

            # 身份映射
            same_x = self.gen_F(real_x, training=True)
            same_y = self.gen_G(real_y, training=True)

            # 判别器输出
            disc_real_x = self.disc_X(real_x, training=True)
            disc_fake_x = self.disc_X(fake_x, training=True)

            disc_real_y = self.disc_Y(real_y, training=True)
            disc_fake_y = self.disc_Y(fake_y, training=True)

            # 生成器对抗损失
            gen_G_loss = self.generator_loss_fn(disc_fake_y)
            gen_F_loss = self.generator_loss_fn(disc_fake_x)

            # 生成器循环损失
            cycle_loss_G = self.cycle_loss_fn(real_y, cycled_y) * self.lambda_cycle
            cycle_loss_F = self.cycle_loss_fn(real_x, cycled_x) * self.lambda_cycle

            # 生成器身份损失
            id_loss_G = (
                self.identity_loss_fn(real_y, same_y)
                * self.lambda_cycle
                * self.lambda_identity
            )
            id_loss_F = (
                self.identity_loss_fn(real_x, same_x)
                * self.lambda_cycle
                * self.lambda_identity
            )

            # 总生成器损失
            total_loss_G = gen_G_loss + cycle_loss_G + id_loss_G
            total_loss_F = gen_F_loss + cycle_loss_F + id_loss_F

            # 判别器损失
            disc_X_loss = self.discriminator_loss_fn(disc_real_x, disc_fake_x)
            disc_Y_loss = self.discriminator_loss_fn(disc_real_y, disc_fake_y)

        # 获取生成器的梯度
        grads_G = tape.gradient(total_loss_G, self.gen_G.trainable_variables)
        grads_F = tape.gradient(total_loss_F, self.gen_F.trainable_variables)

        # 获取判别器的梯度
        disc_X_grads = tape.gradient(disc_X_loss, self.disc_X.trainable_variables)
        disc_Y_grads = tape.gradient(disc_Y_loss, self.disc_Y.trainable_variables)

        # 更新生成器的权重
        self.gen_G_optimizer.apply_gradients(
            zip(grads_G, self.gen_G.trainable_variables)
        )
        self.gen_F_optimizer.apply_gradients(
            zip(grads_F, self.gen_F.trainable_variables)
        )

        # 更新判别器的权重
        self.disc_X_optimizer.apply_gradients(
            zip(disc_X_grads, self.disc_X.trainable_variables)
        )
        self.disc_Y_optimizer.apply_gradients(
            zip(disc_Y_grads, self.disc_Y.trainable_variables)
        )

        return {
            "G_loss": total_loss_G,
            "F_loss": total_loss_F,
            "D_X_loss": disc_X_loss,
            "D_Y_loss": disc_Y_loss,
        }

class GANMonitor(keras.callbacks.Callback):
    """一个回调函数，用于在每个纪元后生成和保存图像"""

    def __init__(self, num_img=4):
        self.num_img = num_img

    def on_epoch_end(self, epoch, logs=None):
        _, ax = plt.subplots(4, 2, figsize=(12, 12))
        for i, img in enumerate(test_horses.take(self.num_img)):
            prediction = self.model.gen_G(img)[0].numpy()
            prediction = (prediction * 127.5 + 127.5).astype(np.uint8)
            img = (img[0] * 127.5 + 127.5).numpy().astype(np.uint8)

            ax[i, 0].imshow(img)
            ax[i, 1].imshow(prediction)
            ax[i, 0].set_title("输入图像")
            ax[i, 1].set_title("转换图像")
            ax[i, 0].axis("off")
            ax[i, 1].axis("off")

            prediction = keras.utils.array_to_img(prediction)
            prediction.save(
                "generated_img_{i}_{epoch}.png".format(i=i, epoch=epoch + 1)
            )
        plt.show()
        plt.close()

# 用于评估对抗损失的损失函数
adv_loss_fn = keras.losses.MeanSquaredError()

# 定义生成器的损失函数
def generator_loss_fn(fake):
    fake_loss = adv_loss_fn(tf.ones_like(fake), fake)
    return fake_loss


# 定义鉴别器的损失函数
def discriminator_loss_fn(real, fake):
    real_loss = adv_loss_fn(tf.ones_like(real), real)
    fake_loss = adv_loss_fn(tf.zeros_like(fake), fake)
    return (real_loss + fake_loss) * 0.5


# 创建循环生成对抗网络模型
cycle_gan_model = CycleGan(
    generator_G=gen_G, generator_F=gen_F, discriminator_X=disc_X, discriminator_Y=disc_Y
)

# 编译模型
cycle_gan_model.compile(
    gen_G_optimizer=keras.optimizers.legacy.Adam(learning_rate=2e-4, beta_1=0.5),
    gen_F_optimizer=keras.optimizers.legacy.Adam(learning_rate=2e-4, beta_1=0.5),
    disc_X_optimizer=keras.optimizers.legacy.Adam(learning_rate=2e-4, beta_1=0.5),
    disc_Y_optimizer=keras.optimizers.legacy.Adam(learning_rate=2e-4, beta_1=0.5),
    gen_loss_fn=generator_loss_fn,
    disc_loss_fn=discriminator_loss_fn,
)
# 回调
plotter = GANMonitor()
checkpoint_filepath = "./model_checkpoints/cyclegan_checkpoints.{epoch:03d}"
model_checkpoint_callback = keras.callbacks.ModelCheckpoint(
    filepath=checkpoint_filepath, save_weights_only=True
)

# 这里我们只训练一个纪元，因为每个纪元大约需要
# 7分钟在单个P100机器上。
cycle_gan_model.fit(
    tf.data.Dataset.zip((train_horses, train_zebras)),
    epochs=1,
    callbacks=[plotter, model_checkpoint_callback],
)

# 该模型训练了90个纪元。我们将在这里加载这些权重
# 一旦权重加载完成，我们将从测试
# 数据中提取几个样本并检查模型的性能。

!curl -LO https://github.com/AakashKumarNain/CycleGAN_TF2/releases/download/v1.0/saved_checkpoints.zip
!unzip -qq saved_checkpoints.zip

# 加载检查点
weight_file = "./saved_checkpoints/cyclegan_checkpoints.090"
cycle_gan_model.load_weights(weight_file).expect_partial()
print("权重加载成功")

_, ax = plt.subplots(4, 2, figsize=(10, 15))
for i, img in enumerate(test_horses.take(4)):
    prediction = cycle_gan_model.gen_G(img, training=False)[0].numpy()
    prediction = (prediction * 127.5 + 127.5).astype(np.uint8)
    img = (img[0] * 127.5 + 127.5).numpy().astype(np.uint8)

    ax[i, 0].imshow(img)
    ax[i, 1].imshow(prediction)
    ax[i, 0].set_title("输入图像")
    ax[i, 0].set_title("输入图像")
    ax[i, 1].set_title("转换图像")
    ax[i, 0].axis("off")
    ax[i, 1].axis("off")

    prediction = keras.utils.array_to_img(prediction)
    prediction.save("predicted_img_{i}.png".format(i=i))
plt.tight_layout()
plt.show()

  % 总计    % 接收 % 传输  平均速度   时间    时间     时间  当前
                                 下载  上传   总计   消耗    剩余  速度
100   634  100   634    0     0   2874      0 --:--:-- --:--:-- --:--:--  2881
100  273M  100  273M    0     0  1736k      0  0:02:41  0:02:41 --:--:-- 2049k

权重加载成功