内容简介：介绍CGAN和ACGAN的原理，通过引入额外的Condition来控制生成的图片，并在DCGAN和WGAN的基础上进行实现样本x可以包含一些属性，或者说条件，记作y例如MNIST中每张图片对应的数字可以是0至9
介绍CGAN和ACGAN的原理，通过引入额外的Condition来控制生成的图片，并在DCGAN和WGAN的基础上进行实现
CGAN原理
样本x可以包含一些属性，或者说条件，记作y
例如MNIST中每张图片对应的数字可以是0至9
从一张图来了解CGAN（Conditional GAN）的思想
生成器G从随机噪音z和条件y生成假样本，判别器D接受真假样本和条件y，判断样本是否为满足条件y的真实样本
总的目标函数如下
实现
先用MNIST，在DCGAN的基础上稍作改动以实现CGAN
加载库
# -*- coding: utf-8 -*- import tensorflow as tf import numpy as np import matplotlib.pyplot as plt %matplotlib inline import os, imageio from tqdm import tqdm 复制代码
加载数据，指定 one_hot=True
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
复制代码
定义一些常量、网络输入、辅助函数，这里加上了 y_label 和 y_noise
batch_size = 100
z_dim = 100
WIDTH = 28
HEIGHT = 28
LABEL = 10
OUTPUT_DIR = 'samples'
if not os.path.exists(OUTPUT_DIR):
os.mkdir(OUTPUT_DIR)
X = tf.placeholder(dtype=tf.float32, shape=[None, HEIGHT, WIDTH, 1], name='X')
y_label = tf.placeholder(dtype=tf.float32, shape=[None, HEIGHT, WIDTH, LABEL], name='y_label')
noise = tf.placeholder(dtype=tf.float32, shape=[None, z_dim], name='noise')
y_noise = tf.placeholder(dtype=tf.float32, shape=[None, LABEL], name='y_noise')
is_training = tf.placeholder(dtype=tf.bool, name='is_training')
def lrelu(x, leak=0.2):
return tf.maximum(x, leak * x)
def sigmoid_cross_entropy_with_logits(x, y):
return tf.nn.sigmoid_cross_entropy_with_logits(logits=x, labels=y)
复制代码
判别器部分
def discriminator(image, label, reuse=None, is_training=is_training):
momentum = 0.9
with tf.variable_scope('discriminator', reuse=reuse):
h0 = tf.concat([image, label], axis=3)
h0 = lrelu(tf.layers.conv2d(h0, kernel_size=5, filters=64, strides=2, padding='same'))
h1 = tf.layers.conv2d(h0, kernel_size=5, filters=128, strides=2, padding='same')
h1 = lrelu(tf.contrib.layers.batch_norm(h1, is_training=is_training, decay=momentum))
h2 = tf.layers.conv2d(h1, kernel_size=5, filters=256, strides=2, padding='same')
h2 = lrelu(tf.contrib.layers.batch_norm(h2, is_training=is_training, decay=momentum))
h3 = tf.layers.conv2d(h2, kernel_size=5, filters=512, strides=2, padding='same')
h3 = lrelu(tf.contrib.layers.batch_norm(h3, is_training=is_training, decay=momentum))
h4 = tf.contrib.layers.flatten(h3)
h4 = tf.layers.dense(h4, units=1)
return tf.nn.sigmoid(h4), h4
复制代码
生成器部分
def generator(z, label, is_training=is_training):
momentum = 0.9
with tf.variable_scope('generator', reuse=None):
d = 3
z = tf.concat([z, label], axis=1)
h0 = tf.layers.dense(z, units=d * d * 512)
h0 = tf.reshape(h0, shape=[-1, d, d, 512])
h0 = tf.nn.relu(tf.contrib.layers.batch_norm(h0, is_training=is_training, decay=momentum))
h1 = tf.layers.conv2d_transpose(h0, kernel_size=5, filters=256, strides=2, padding='same')
h1 = tf.nn.relu(tf.contrib.layers.batch_norm(h1, is_training=is_training, decay=momentum))
h2 = tf.layers.conv2d_transpose(h1, kernel_size=5, filters=128, strides=2, padding='same')
h2 = tf.nn.relu(tf.contrib.layers.batch_norm(h2, is_training=is_training, decay=momentum))
h3 = tf.layers.conv2d_transpose(h2, kernel_size=5, filters=64, strides=2, padding='same')
h3 = tf.nn.relu(tf.contrib.layers.batch_norm(h3, is_training=is_training, decay=momentum))
h4 = tf.layers.conv2d_transpose(h3, kernel_size=5, filters=1, strides=1, padding='valid', activation=tf.nn.tanh, name='g')
return h4
复制代码
损失函数
g = generator(noise, y_noise)
d_real, d_real_logits = discriminator(X, y_label)
d_fake, d_fake_logits = discriminator(g, y_label, reuse=True)
vars_g = [var for var in tf.trainable_variables() if var.name.startswith('generator')]
vars_d = [var for var in tf.trainable_variables() if var.name.startswith('discriminator')]
loss_d_real = tf.reduce_mean(sigmoid_cross_entropy_with_logits(d_real_logits, tf.ones_like(d_real)))
loss_d_fake = tf.reduce_mean(sigmoid_cross_entropy_with_logits(d_fake_logits, tf.zeros_like(d_fake)))
loss_g = tf.reduce_mean(sigmoid_cross_entropy_with_logits(d_fake_logits, tf.ones_like(d_fake)))
loss_d = loss_d_real + loss_d_fake
复制代码
优化函数
update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
with tf.control_dependencies(update_ops):
optimizer_d = tf.train.AdamOptimizer(learning_rate=0.0002, beta1=0.5).minimize(loss_d, var_list=vars_d)
optimizer_g = tf.train.AdamOptimizer(learning_rate=0.0002, beta1=0.5).minimize(loss_g, var_list=vars_g)
复制代码
拼接图片的函数
def montage(images):
if isinstance(images, list):
images = np.array(images)
img_h = images.shape[1]
img_w = images.shape[2]
n_plots = int(np.ceil(np.sqrt(images.shape[0])))
m = np.ones((images.shape[1] * n_plots + n_plots + 1, images.shape[2] * n_plots + n_plots + 1)) * 0.5
for i in range(n_plots):
for j in range(n_plots):
this_filter = i * n_plots + j
if this_filter < images.shape[0]:
this_img = images[this_filter]
m[1 + i + i * img_h:1 + i + (i + 1) * img_h,
1 + j + j * img_w:1 + j + (j + 1) * img_w] = this_img
return m
复制代码
训练模型，加入条件信息
sess = tf.Session()
sess.run(tf.global_variables_initializer())
z_samples = np.random.uniform(-1.0, 1.0, [batch_size, z_dim]).astype(np.float32)
y_samples = np.zeros([batch_size, LABEL])
for i in range(LABEL):
for j in range(LABEL):
y_samples[i * LABEL + j, i] = 1
samples = []
loss = {'d': [], 'g': []}
for i in tqdm(range(60000)):
n = np.random.uniform(-1.0, 1.0, [batch_size, z_dim]).astype(np.float32)
batch, label = mnist.train.next_batch(batch_size=batch_size)
batch = np.reshape(batch, [batch_size, HEIGHT, WIDTH, 1])
batch = (batch - 0.5) * 2
yn = np.copy(label)
yl = np.reshape(label, [batch_size, 1, 1, LABEL])
yl = yl * np.ones([batch_size, HEIGHT, WIDTH, LABEL])
d_ls, g_ls = sess.run([loss_d, loss_g], feed_dict={X: batch, noise: n, y_label: yl, y_noise: yn, is_training: True})
loss['d'].append(d_ls)
loss['g'].append(g_ls)
sess.run(optimizer_d, feed_dict={X: batch, noise: n, y_label: yl, y_noise: yn, is_training: True})
sess.run(optimizer_g, feed_dict={X: batch, noise: n, y_label: yl, y_noise: yn, is_training: True})
sess.run(optimizer_g, feed_dict={X: batch, noise: n, y_label: yl, y_noise: yn, is_training: True})
if i % 1000 == 0:
print(i, d_ls, g_ls)
gen_imgs = sess.run(g, feed_dict={noise: z_samples, y_noise: y_samples, is_training: False})
gen_imgs = (gen_imgs + 1) / 2
imgs = [img[:, :, 0] for img in gen_imgs]
gen_imgs = montage(imgs)
plt.axis('off')
plt.imshow(gen_imgs, cmap='gray')
imageio.imsave(os.path.join(OUTPUT_DIR, 'sample_%d.jpg' % i), gen_imgs)
plt.show()
samples.append(gen_imgs)
plt.plot(loss['d'], label='Discriminator')
plt.plot(loss['g'], label='Generator')
plt.legend(loc='upper right')
plt.savefig('Loss.png')
plt.show()
imageio.mimsave(os.path.join(OUTPUT_DIR, 'samples.gif'), samples, fps=5)
复制代码
生成的手写数字图片如下，每一行对应的数字相同
保存模型，便于后续使用
saver = tf.train.Saver() saver.save(sess, './mnist_cgan', global_step=60000) 复制代码
在单机上使用模型生成手写数字图片
# -*- coding: utf-8 -*-
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
batch_size = 100
z_dim = 100
LABEL = 10
def montage(images):
if isinstance(images, list):
images = np.array(images)
img_h = images.shape[1]
img_w = images.shape[2]
n_plots = int(np.ceil(np.sqrt(images.shape[0])))
m = np.ones((images.shape[1] * n_plots + n_plots + 1, images.shape[2] * n_plots + n_plots + 1)) * 0.5
for i in range(n_plots):
for j in range(n_plots):
this_filter = i * n_plots + j
if this_filter < images.shape[0]:
this_img = images[this_filter]
m[1 + i + i * img_h:1 + i + (i + 1) * img_h,
1 + j + j * img_w:1 + j + (j + 1) * img_w] = this_img
return m
sess = tf.Session()
sess.run(tf.global_variables_initializer())
saver = tf.train.import_meta_graph('./mnist_cgan-60000.meta')
saver.restore(sess, tf.train.latest_checkpoint('./'))
graph = tf.get_default_graph()
g = graph.get_tensor_by_name('generator/g/Tanh:0')
noise = graph.get_tensor_by_name('noise:0')
y_noise = graph.get_tensor_by_name('y_noise:0')
is_training = graph.get_tensor_by_name('is_training:0')
n = np.random.uniform(-1.0, 1.0, [batch_size, z_dim]).astype(np.float32)
y_samples = np.zeros([batch_size, LABEL])
for i in range(LABEL):
for j in range(LABEL):
y_samples[i * LABEL + j, i] = 1
gen_imgs = sess.run(g, feed_dict={noise: n, y_noise: y_samples, is_training: False})
gen_imgs = (gen_imgs + 1) / 2
imgs = [img[:, :, 0] for img in gen_imgs]
gen_imgs = montage(imgs)
plt.axis('off')
plt.imshow(gen_imgs, cmap='gray')
plt.show()
复制代码
讲条件的CelebA
了解CGAN的原理和实现之后，再尝试下别的数据集，比如之前用过的CelebA
CelebA提供了每张图片40个属性的01标注，这里将Male（是否为男性）作为条件，在WGAN的基础上实现CGAN
加载库
# -*- coding: utf-8 -*- import tensorflow as tf import numpy as np import os import matplotlib.pyplot as plt %matplotlib inline from imageio import imread, imsave, mimsave import cv2 import glob from tqdm import tqdm 复制代码
加载图片
images = glob.glob('celeba/*.jpg')
print(len(images))
复制代码
读取图片的Male标签
tags = {}
target = 'Male'
with open('list_attr_celeba.txt', 'r') as fr:
lines = fr.readlines()
all_tags = lines[0].strip('\n').split()
for i in range(1, len(lines)):
line = lines[i].strip('\n').split()
if int(line[all_tags.index(target) + 1]) == 1:
tags[line[0]] = [1, 0] # 男
else:
tags[line[0]] = [0, 1] # 女
print(len(tags))
print(all_tags)
复制代码
定义一些常量、网络输入、辅助函数
batch_size = 100
z_dim = 100
WIDTH = 64
HEIGHT = 64
LABEL = 2
LAMBDA = 10
DIS_ITERS = 3 # 5
OUTPUT_DIR = 'samples'
if not os.path.exists(OUTPUT_DIR):
os.mkdir(OUTPUT_DIR)
X = tf.placeholder(dtype=tf.float32, shape=[batch_size, HEIGHT, WIDTH, 3], name='X')
y_label = tf.placeholder(dtype=tf.float32, shape=[batch_size, HEIGHT, WIDTH, LABEL], name='y_label')
noise = tf.placeholder(dtype=tf.float32, shape=[batch_size, z_dim], name='noise')
y_noise = tf.placeholder(dtype=tf.float32, shape=[batch_size, LABEL], name='y_noise')
is_training = tf.placeholder(dtype=tf.bool, name='is_training')
def lrelu(x, leak=0.2):
return tf.maximum(x, leak * x)
复制代码
判别器部分
def discriminator(image, label, reuse=None, is_training=is_training):
momentum = 0.9
with tf.variable_scope('discriminator', reuse=reuse):
h0 = tf.concat([image, label], axis=3)
h0 = lrelu(tf.layers.conv2d(h0, kernel_size=5, filters=64, strides=2, padding='same'))
h1 = lrelu(tf.layers.conv2d(h0, kernel_size=5, filters=128, strides=2, padding='same'))
h2 = lrelu(tf.layers.conv2d(h1, kernel_size=5, filters=256, strides=2, padding='same'))
h3 = lrelu(tf.layers.conv2d(h2, kernel_size=5, filters=512, strides=2, padding='same'))
h4 = tf.contrib.layers.flatten(h3)
h4 = tf.layers.dense(h4, units=1)
return h4
复制代码
生成器部分
def generator(z, label, is_training=is_training):
momentum = 0.9
with tf.variable_scope('generator', reuse=None):
d = 4
z = tf.concat([z, label], axis=1)
h0 = tf.layers.dense(z, units=d * d * 512)
h0 = tf.reshape(h0, shape=[-1, d, d, 512])
h0 = tf.nn.relu(tf.contrib.layers.batch_norm(h0, is_training=is_training, decay=momentum))
h1 = tf.layers.conv2d_transpose(h0, kernel_size=5, filters=256, strides=2, padding='same')
h1 = tf.nn.relu(tf.contrib.layers.batch_norm(h1, is_training=is_training, decay=momentum))
h2 = tf.layers.conv2d_transpose(h1, kernel_size=5, filters=128, strides=2, padding='same')
h2 = tf.nn.relu(tf.contrib.layers.batch_norm(h2, is_training=is_training, decay=momentum))
h3 = tf.layers.conv2d_transpose(h2, kernel_size=5, filters=64, strides=2, padding='same')
h3 = tf.nn.relu(tf.contrib.layers.batch_norm(h3, is_training=is_training, decay=momentum))
h4 = tf.layers.conv2d_transpose(h3, kernel_size=5, filters=3, strides=2, padding='same', activation=tf.nn.tanh, name='g')
return h4
复制代码
定义损失函数
g = generator(noise, y_noise)
d_real = discriminator(X, y_label)
d_fake = discriminator(g, y_label, reuse=True)
loss_d_real = -tf.reduce_mean(d_real)
loss_d_fake = tf.reduce_mean(d_fake)
loss_g = -tf.reduce_mean(d_fake)
loss_d = loss_d_real + loss_d_fake
alpha = tf.random_uniform(shape=[batch_size, 1, 1, 1], minval=0., maxval=1.)
interpolates = alpha * X + (1 - alpha) * g
grad = tf.gradients(discriminator(interpolates, y_label, reuse=True), [interpolates])[0]
slop = tf.sqrt(tf.reduce_sum(tf.square(grad), axis=[1]))
gp = tf.reduce_mean((slop - 1.) ** 2)
loss_d += LAMBDA * gp
vars_g = [var for var in tf.trainable_variables() if var.name.startswith('generator')]
vars_d = [var for var in tf.trainable_variables() if var.name.startswith('discriminator')]
复制代码
定义优化器
update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
with tf.control_dependencies(update_ops):
optimizer_d = tf.train.AdamOptimizer(learning_rate=0.0002, beta1=0.5).minimize(loss_d, var_list=vars_d)
optimizer_g = tf.train.AdamOptimizer(learning_rate=0.0002, beta1=0.5).minimize(loss_g, var_list=vars_g)
复制代码
拼接图片的函数
def montage(images):
if isinstance(images, list):
images = np.array(images)
img_h = images.shape[1]
img_w = images.shape[2]
n_plots = int(np.ceil(np.sqrt(images.shape[0])))
if len(images.shape) == 4 and images.shape[3] == 3:
m = np.ones(
(images.shape[1] * n_plots + n_plots + 1,
images.shape[2] * n_plots + n_plots + 1, 3)) * 0.5
elif len(images.shape) == 4 and images.shape[3] == 1:
m = np.ones(
(images.shape[1] * n_plots + n_plots + 1,
images.shape[2] * n_plots + n_plots + 1, 1)) * 0.5
elif len(images.shape) == 3:
m = np.ones(
(images.shape[1] * n_plots + n_plots + 1,
images.shape[2] * n_plots + n_plots + 1)) * 0.5
else:
raise ValueError('Could not parse image shape of {}'.format(images.shape))
for i in range(n_plots):
for j in range(n_plots):
this_filter = i * n_plots + j
if this_filter < images.shape[0]:
this_img = images[this_filter]
m[1 + i + i * img_h:1 + i + (i + 1) * img_h,
1 + j + j * img_w:1 + j + (j + 1) * img_w] = this_img
return m
复制代码
整理数据
X_all = []
Y_all = []
for i in tqdm(range(len(images))):
image = imread(images[i])
h = image.shape[0]
w = image.shape[1]
if h > w:
image = image[h // 2 - w // 2: h // 2 + w // 2, :, :]
else:
image = image[:, w // 2 - h // 2: w // 2 + h // 2, :]
image = cv2.resize(image, (WIDTH, HEIGHT))
image = (image / 255. - 0.5) * 2
X_all.append(image)
image_name = images[i][images[i].find('/') + 1:]
Y_all.append(tags[image_name])
X_all = np.array(X_all)
Y_all = np.array(Y_all)
print(X_all.shape, Y_all.shape)
复制代码
查看数据样例
for i in range(10):
plt.imshow((X_all[i, :, :, :] + 1) / 2)
plt.show()
print(Y_all[i, :])
复制代码
定义随机产生批数据的函数
def get_random_batch():
data_index = np.arange(X_all.shape[0])
np.random.shuffle(data_index)
data_index = data_index[:batch_size]
X_batch = X_all[data_index, :, :, :]
Y_batch = Y_all[data_index, :]
yn = np.copy(Y_batch)
yl = np.reshape(Y_batch, [batch_size, 1, 1, LABEL])
yl = yl * np.ones([batch_size, HEIGHT, WIDTH, LABEL])
return X_batch, yn, yl
复制代码
训练模型
sess = tf.Session()
sess.run(tf.global_variables_initializer())
zs = np.random.uniform(-1.0, 1.0, [batch_size // 2, z_dim]).astype(np.float32)
z_samples = []
y_samples = []
for i in range(batch_size // 2):
z_samples.append(zs[i, :])
y_samples.append([1, 0])
z_samples.append(zs[i, :])
y_samples.append([0, 1])
samples = []
loss = {'d': [], 'g': []}
for i in tqdm(range(60000)):
for j in range(DIS_ITERS):
n = np.random.uniform(-1.0, 1.0, [batch_size, z_dim]).astype(np.float32)
X_batch, yn, yl = get_random_batch()
_, d_ls = sess.run([optimizer_d, loss_d], feed_dict={X: X_batch, noise: n, y_label: yl, y_noise: yn, is_training: True})
_, g_ls = sess.run([optimizer_g, loss_g], feed_dict={X: X_batch, noise: n, y_label: yl, y_noise: yn, is_training: True})
loss['d'].append(d_ls)
loss['g'].append(g_ls)
if i % 500 == 0:
print(i, d_ls, g_ls)
gen_imgs = sess.run(g, feed_dict={noise: z_samples, y_noise: y_samples, is_training: False})
gen_imgs = (gen_imgs + 1) / 2
imgs = [img[:, :, :] for img in gen_imgs]
gen_imgs = montage(imgs)
plt.axis('off')
plt.imshow(gen_imgs)
imsave(os.path.join(OUTPUT_DIR, 'sample_%d.jpg' % i), gen_imgs)
plt.show()
samples.append(gen_imgs)
plt.plot(loss['d'], label='Discriminator')
plt.plot(loss['g'], label='Generator')
plt.legend(loc='upper right')
plt.savefig('Loss.png')
plt.show()
mimsave(os.path.join(OUTPUT_DIR, 'samples.gif'), samples, fps=10)
复制代码
结果如下，对于每一组图片，噪音部分相同但条件不同，男左女右
保存模型
saver = tf.train.Saver() saver.save(sess, './celeba_cgan', global_step=60000) 复制代码
ACGAN
再通过一张图了解ACGAN（Auxiliary Classifier GAN）的原理
和CGAN不同的是，C不直接输入D。D不仅需要判断每个样本的真假，还需要完成一个分类任务即预测C，通过增加一个辅助分类器实现
对D而言，损失函数如下
对G而言，损失函数如下
还是以CelebA的Male作为条件，在WGAN的基础上实现ACGAN
加载库
# -*- coding: utf-8 -*- import tensorflow as tf import numpy as np import os import matplotlib.pyplot as plt %matplotlib inline from imageio import imread, imsave, mimsave import cv2 import glob from tqdm import tqdm 复制代码
加载图片
images = glob.glob('celeba/*.jpg')
print(len(images))
复制代码
读取图片的Male标签
tags = {}
target = 'Male'
with open('list_attr_celeba.txt', 'r') as fr:
lines = fr.readlines()
all_tags = lines[0].strip('\n').split()
for i in range(1, len(lines)):
line = lines[i].strip('\n').split()
if int(line[all_tags.index(target) + 1]) == 1:
tags[line[0]] = [1, 0] # 男
else:
tags[line[0]] = [0, 1] # 女
print(len(tags))
print(all_tags)
复制代码
定义一些常量、网络输入、辅助函数
batch_size = 100
z_dim = 100
WIDTH = 64
HEIGHT = 64
LABEL = 2
LAMBDA = 10
DIS_ITERS = 3 # 5
OUTPUT_DIR = 'samples'
if not os.path.exists(OUTPUT_DIR):
os.mkdir(OUTPUT_DIR)
X = tf.placeholder(dtype=tf.float32, shape=[batch_size, HEIGHT, WIDTH, 3], name='X')
Y = tf.placeholder(dtype=tf.float32, shape=[batch_size, LABEL], name='Y')
noise = tf.placeholder(dtype=tf.float32, shape=[batch_size, z_dim], name='noise')
is_training = tf.placeholder(dtype=tf.bool, name='is_training')
def lrelu(x, leak=0.2):
return tf.maximum(x, leak * x)
复制代码
判别器部分，去掉条件输入，增加分类输出
def discriminator(image, reuse=None, is_training=is_training):
momentum = 0.9
with tf.variable_scope('discriminator', reuse=reuse):
h0 = lrelu(tf.layers.conv2d(image, kernel_size=5, filters=64, strides=2, padding='same'))
h1 = lrelu(tf.layers.conv2d(h0, kernel_size=5, filters=128, strides=2, padding='same'))
h2 = lrelu(tf.layers.conv2d(h1, kernel_size=5, filters=256, strides=2, padding='same'))
h3 = lrelu(tf.layers.conv2d(h2, kernel_size=5, filters=512, strides=2, padding='same'))
h4 = tf.contrib.layers.flatten(h3)
Y_ = tf.layers.dense(h4, units=LABEL)
h4 = tf.layers.dense(h4, units=1)
return h4, Y_
复制代码
生成器部分，没有任何改动
def generator(z, label, is_training=is_training):
momentum = 0.9
with tf.variable_scope('generator', reuse=None):
d = 4
z = tf.concat([z, label], axis=1)
h0 = tf.layers.dense(z, units=d * d * 512)
h0 = tf.reshape(h0, shape=[-1, d, d, 512])
h0 = tf.nn.relu(tf.contrib.layers.batch_norm(h0, is_training=is_training, decay=momentum))
h1 = tf.layers.conv2d_transpose(h0, kernel_size=5, filters=256, strides=2, padding='same')
h1 = tf.nn.relu(tf.contrib.layers.batch_norm(h1, is_training=is_training, decay=momentum))
h2 = tf.layers.conv2d_transpose(h1, kernel_size=5, filters=128, strides=2, padding='same')
h2 = tf.nn.relu(tf.contrib.layers.batch_norm(h2, is_training=is_training, decay=momentum))
h3 = tf.layers.conv2d_transpose(h2, kernel_size=5, filters=64, strides=2, padding='same')
h3 = tf.nn.relu(tf.contrib.layers.batch_norm(h3, is_training=is_training, decay=momentum))
h4 = tf.layers.conv2d_transpose(h3, kernel_size=5, filters=3, strides=2, padding='same', activation=tf.nn.tanh, name='g')
return h4
复制代码
定义损失函数，加上分类部分对应的损失。理论上分类问题应该用交叉熵作为损失函数，这里使用MSE效果也不错
g = generator(noise, Y)
d_real, y_real = discriminator(X)
d_fake, y_fake = discriminator(g, reuse=True)
loss_d_real = -tf.reduce_mean(d_real)
loss_d_fake = tf.reduce_mean(d_fake)
loss_cls_real = tf.losses.mean_squared_error(Y, y_real)
loss_cls_fake = tf.losses.mean_squared_error(Y, y_fake)
loss_d = loss_d_real + loss_d_fake + loss_cls_real
loss_g = -tf.reduce_mean(d_fake) + loss_cls_fake
alpha = tf.random_uniform(shape=[batch_size, 1, 1, 1], minval=0., maxval=1.)
interpolates = alpha * X + (1 - alpha) * g
grad = tf.gradients(discriminator(interpolates, reuse=True), [interpolates])[0]
slop = tf.sqrt(tf.reduce_sum(tf.square(grad), axis=[1]))
gp = tf.reduce_mean((slop - 1.) ** 2)
loss_d += LAMBDA * gp
vars_g = [var for var in tf.trainable_variables() if var.name.startswith('generator')]
vars_d = [var for var in tf.trainable_variables() if var.name.startswith('discriminator')]
复制代码
定义优化器
update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
with tf.control_dependencies(update_ops):
optimizer_d = tf.train.AdamOptimizer(learning_rate=0.0002, beta1=0.5).minimize(loss_d, var_list=vars_d)
optimizer_g = tf.train.AdamOptimizer(learning_rate=0.0002, beta1=0.5).minimize(loss_g, var_list=vars_g)
复制代码
拼接图片的函数
def montage(images):
if isinstance(images, list):
images = np.array(images)
img_h = images.shape[1]
img_w = images.shape[2]
n_plots = int(np.ceil(np.sqrt(images.shape[0])))
if len(images.shape) == 4 and images.shape[3] == 3:
m = np.ones(
(images.shape[1] * n_plots + n_plots + 1,
images.shape[2] * n_plots + n_plots + 1, 3)) * 0.5
elif len(images.shape) == 4 and images.shape[3] == 1:
m = np.ones(
(images.shape[1] * n_plots + n_plots + 1,
images.shape[2] * n_plots + n_plots + 1, 1)) * 0.5
elif len(images.shape) == 3:
m = np.ones(
(images.shape[1] * n_plots + n_plots + 1,
images.shape[2] * n_plots + n_plots + 1)) * 0.5
else:
raise ValueError('Could not parse image shape of {}'.format(images.shape))
for i in range(n_plots):
for j in range(n_plots):
this_filter = i * n_plots + j
if this_filter < images.shape[0]:
this_img = images[this_filter]
m[1 + i + i * img_h:1 + i + (i + 1) * img_h,
1 + j + j * img_w:1 + j + (j + 1) * img_w] = this_img
return m
复制代码
整理数据
X_all = []
Y_all = []
for i in tqdm(range(len(images))):
image = imread(images[i])
h = image.shape[0]
w = image.shape[1]
if h > w:
image = image[h // 2 - w // 2: h // 2 + w // 2, :, :]
else:
image = image[:, w // 2 - h // 2: w // 2 + h // 2, :]
image = cv2.resize(image, (WIDTH, HEIGHT))
image = (image / 255. - 0.5) * 2
X_all.append(image)
image_name = images[i][images[i].find('/') + 1:]
Y_all.append(tags[image_name])
X_all = np.array(X_all)
Y_all = np.array(Y_all)
print(X_all.shape, Y_all.shape)
复制代码
查看数据样例
for i in range(10):
plt.imshow((X_all[i, :, :, :] + 1) / 2)
plt.show()
print(Y_all[i, :])
复制代码
定义随机产生批数据的函数
def get_random_batch():
data_index = np.arange(X_all.shape[0])
np.random.shuffle(data_index)
data_index = data_index[:batch_size]
X_batch = X_all[data_index, :, :, :]
Y_batch = Y_all[data_index, :]
return X_batch, Y_batch
复制代码
训练模型，根据ACGAN作相应调整
sess = tf.Session()
sess.run(tf.global_variables_initializer())
zs = np.random.uniform(-1.0, 1.0, [batch_size // 2, z_dim]).astype(np.float32)
z_samples = []
y_samples = []
for i in range(batch_size // 2):
z_samples.append(zs[i, :])
y_samples.append([1, 0])
z_samples.append(zs[i, :])
y_samples.append([0, 1])
samples = []
loss = {'d': [], 'g': []}
for i in tqdm(range(60000)):
for j in range(DIS_ITERS):
n = np.random.uniform(-1.0, 1.0, [batch_size, z_dim]).astype(np.float32)
X_batch, Y_batch = get_random_batch()
_, d_ls = sess.run([optimizer_d, loss_d], feed_dict={X: X_batch, Y: Y_batch, noise: n, is_training: True})
_, g_ls = sess.run([optimizer_g, loss_g], feed_dict={X: X_batch, Y: Y_batch, noise: n, is_training: True})
loss['d'].append(d_ls)
loss['g'].append(g_ls)
if i % 500 == 0:
print(i, d_ls, g_ls)
gen_imgs = sess.run(g, feed_dict={noise: z_samples, Y: y_samples, is_training: False})
gen_imgs = (gen_imgs + 1) / 2
imgs = [img[:, :, :] for img in gen_imgs]
gen_imgs = montage(imgs)
plt.axis('off')
plt.imshow(gen_imgs)
imsave(os.path.join(OUTPUT_DIR, 'sample_%d.jpg' % i), gen_imgs)
plt.show()
samples.append(gen_imgs)
plt.plot(loss['d'], label='Discriminator')
plt.plot(loss['g'], label='Generator')
plt.legend(loc='upper right')
plt.savefig('Loss.png')
plt.show()
mimsave(os.path.join(OUTPUT_DIR, 'samples.gif'), samples, fps=10)
复制代码
结果如下，比CGAN的结果好一些，崩掉的情况比较少，而且人脸更真实更清晰
保存模型
saver = tf.train.Saver() saver.save(sess, './celeba_acgan', global_step=60000) 复制代码
在单机上加载模型，进行以下两个尝试：
- 固定噪音，渐变条件;
- 固定条件，渐变噪音。
# -*- coding: utf-8 -*-
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
batch_size = 100
z_dim = 100
LABEL = 2
def montage(images):
if isinstance(images, list):
images = np.array(images)
img_h = images.shape[1]
img_w = images.shape[2]
n_plots = int(np.ceil(np.sqrt(images.shape[0])))
if len(images.shape) == 4 and images.shape[3] == 3:
m = np.ones(
(images.shape[1] * n_plots + n_plots + 1,
images.shape[2] * n_plots + n_plots + 1, 3)) * 0.5
elif len(images.shape) == 4 and images.shape[3] == 1:
m = np.ones(
(images.shape[1] * n_plots + n_plots + 1,
images.shape[2] * n_plots + n_plots + 1, 1)) * 0.5
elif len(images.shape) == 3:
m = np.ones(
(images.shape[1] * n_plots + n_plots + 1,
images.shape[2] * n_plots + n_plots + 1)) * 0.5
else:
raise ValueError('Could not parse image shape of {}'.format(images.shape))
for i in range(n_plots):
for j in range(n_plots):
this_filter = i * n_plots + j
if this_filter < images.shape[0]:
this_img = images[this_filter]
m[1 + i + i * img_h:1 + i + (i + 1) * img_h,
1 + j + j * img_w:1 + j + (j + 1) * img_w] = this_img
return m
sess = tf.Session()
sess.run(tf.global_variables_initializer())
saver = tf.train.import_meta_graph('./celeba_acgan-60000.meta')
saver.restore(sess, tf.train.latest_checkpoint('./'))
graph = tf.get_default_graph()
g = graph.get_tensor_by_name('generator/g/Tanh:0')
noise = graph.get_tensor_by_name('noise:0')
Y = graph.get_tensor_by_name('Y:0')
is_training = graph.get_tensor_by_name('is_training:0')
# 固定噪音，渐变条件
n = np.random.uniform(-1.0, 1.0, [10, z_dim]).astype(np.float32)
ns = []
y_samples = []
for i in range(100):
ns.append(n[i // 10, :])
y_samples.append([i % 10 / 9, 1 - i % 10 / 9])
gen_imgs = sess.run(g, feed_dict={noise: ns, Y: y_samples, is_training: False})
gen_imgs = (gen_imgs + 1) / 2
imgs = [img[:, :, :] for img in gen_imgs]
gen_imgs = montage(imgs)
gen_imgs = np.clip(gen_imgs, 0, 1)
plt.figure(figsize=(8, 8))
plt.axis('off')
plt.imshow(gen_imgs)
plt.show()
# 固定条件，渐变噪音
n = np.random.uniform(-1.0, 1.0, [5, 2, z_dim]).astype(np.float32)
ns = []
y_samples = []
for i in range(5):
for k in range(2):
for j in range(10):
start = n[i, 0, :]
end = n[i, 1, :]
ns.append(start + j * (end - start) / 9)
if k == 0:
y_samples.append([0, 1])
else:
y_samples.append([1, 0])
gen_imgs = sess.run(g, feed_dict={noise: ns, Y: y_samples, is_training: False})
gen_imgs = (gen_imgs + 1) / 2
imgs = [img[:, :, :] for img in gen_imgs]
gen_imgs = montage(imgs)
gen_imgs = np.clip(gen_imgs, 0, 1)
plt.figure(figsize=(8, 8))
plt.axis('off')
plt.imshow(gen_imgs)
plt.show()
复制代码
由女变男的过程
人脸两两之间的渐变
以上就是本文的全部内容，希望本文的内容对大家的学习或者工作能带来一定的帮助，也希望大家多多支持 码农网
猜你喜欢:
- 沈向洋：从深度学习到深度理解
- 深度重建：基于深度学习的图像重建
- 深度网络揭秘之深度网络背后的数学
- 深度解析Python深度学习框架的对比
- 【深度好文】深度分析如何获取方法参数名
- 直观理解深度学习基本概念（小白入门深度学习）
本站部分资源来源于网络，本站转载出于传递更多信息之目的，版权归原作者或者来源机构所有，如转载稿涉及版权问题，请联系我们。
Django 1.0 Template Development
Scott Newman / Packt / 2008 / 24.99
Django is a high-level Python web application framework designed to support the rapid development of dynamic websites, web applications, and web services. Getting the most out of its template system a......一起来看看 《Django 1.0 Template Development》 这本书的介绍吧!
