import tensorflow as tf
import scipy.io as sio
import numpy as np
from PIL import Image
import os
import matplotlib.pyplot as plt
np.set_printoptions(threshold=np.inf) #输出全部矩阵不带省略号
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets('MNIST_data',one_hot=True)
data = mnist.train.images
data = data.reshape(-1,28,28,1)
print(data.shape)
label = mnist.train.labels
print(label.shape)
###########################################################################
width = 28
height = 28
channel = 1
n_class = 10
GAN_type = "SNGAN" # DCGAN, WGAN, WGAN-GP, SNGAN, LSGAN, RSGAN, RaSGAN
batch_size = 128
epochs = 500
epsilon = 1e-14#if epsilon is too big, training of DCGAN is failure.
def deconv(inputs, shape, strides, out_shape, is_sn=False, padding="SAME"):
filters = tf.get_variable("kernel", shape=shape, initializer=tf.random_normal_initializer(stddev=0.02))
bias = tf.get_variable("bias", shape=[shape[-2]], initializer=tf.constant_initializer([0]))
if is_sn:
return tf.nn.conv2d_transpose(inputs, spectral_norm("sn", filters), out_shape, strides, padding) + bias
else:
return tf.nn.conv2d_transpose(inputs, filters, out_shape, strides, padding) + bias
def conv(inputs, shape, strides, is_sn=False, padding="SAME"):
filters = tf.get_variable("kernel", shape=shape, initializer=tf.random_normal_initializer(stddev=0.02))
bias = tf.get_variable("bias", shape=[shape[-1]], initializer=tf.constant_initializer([0]))
if is_sn:
return tf.nn.conv2d(inputs, spectral_norm("sn", filters), strides, padding) + bias
else:
return tf.nn.conv2d(inputs, filters, strides, padding) + bias
def fully_connected(inputs, num_out, is_sn=False):
W = tf.get_variable("W", [inputs.shape[-1], num_out], initializer=tf.random_normal_initializer(stddev=0.02))
b = tf.get_variable("b", [num_out], initializer=tf.constant_initializer([0]))
if is_sn:
return tf.matmul(inputs, spectral_norm("sn", W)) + b
else:
return tf.matmul(inputs, W) + b
def leaky_relu(inputs, slope=0.2):
return tf.maximum(slope*inputs, inputs)
def mapping(x):
max = np.max(x)
min = np.min(x)
return (x - min) * 255.0 / (max - min + epsilon)
def sample_images(self, gen_data,epoch):
gen_data = gen_data[0:64]
r, c = 8, 8
fig, axs = plt.subplots(r, c)
cnt = 0
for i in range(r):
for j in range(c):
axs[i,j].imshow(gen_data[cnt, :,:,0], cmap='gray')
axs[i,j].axis('off')
cnt += 1
if not os.path.exists("c_mnist_result"):
os.makedirs("c_mnist_result")
fig.savefig("c_mnist_result/%d.png" % epoch)
plt.close()
def spectral_norm(name, w, iteration=1):
#Spectral normalization which was published on ICLR2018,please refer to "https://www.researchgate.net/publication/318572189_Spectral_Normalization_for_Generative_Adversarial_Networks"
#This function spectral_norm is forked from "https://github.com/taki0112/Spectral_Normalization-Tensorflow"
w_shape = w.shape.as_list()
w = tf.reshape(w, [-1, w_shape[-1]])
with tf.variable_scope(name, reuse=False):
u = tf.get_variable("u", [1, w_shape[-1]], initializer=tf.truncated_normal_initializer(), trainable=False)
u_hat = u
v_hat = None
def l2_norm(v, eps=1e-12):
return v / (tf.reduce_sum(v ** 2) ** 0.5 + eps)
for i in range(iteration):
v_ = tf.matmul(u_hat, tf.transpose(w))
v_hat = l2_norm(v_)
u_ = tf.matmul(v_hat, w)
u_hat = l2_norm(u_)
sigma = tf.matmul(tf.matmul(v_hat, w), tf.transpose(u_hat))
w_norm = w / sigma
with tf.control_dependencies([u.assign(u_hat)]):
w_norm = tf.reshape(w_norm, w_shape)
return w_norm
def bn(inputs):
mean, var = tf.nn.moments(inputs, axes=[1, 2], keep_dims=True)
scale = tf.get_variable("scale", shape=mean.shape, initializer=tf.constant_initializer([1.0]))
shift = tf.get_variable("shift", shape=mean.shape, initializer=tf.constant_initializer([0.0]))
return (inputs - mean) * scale / (tf.sqrt(var + epsilon)) + shift
def conv_cond_concat(x, y):
"""Concatenate conditioning vector on feature map axis."""
x_shapes = x.get_shape()
y_shapes = y.get_shape()
return tf.concat([
x, y*tf.ones([x_shapes[0], x_shapes[1], x_shapes[2], y_shapes[3]])], 3)
class Generator:
def __init__(self, name):
self.name = name
def __call__(self, Z, label, reuse=False):
with tf.variable_scope(name_or_scope=self.name, reuse=reuse):
# linear
Z = tf.concat([Z,label], 1)
print("g_inputs:", Z.shape)
label_ = tf.reshape(label,[batch_size, 1, 1, n_class])
with tf.variable_scope(name_or_scope="linear"):
output = fully_connected(Z, 4*4*512)
output = tf.nn.relu(output)
output = tf.reshape(output, [batch_size, 4, 4, 512])
output = conv_cond_concat(output,label_)
print("g_fc:", output)
# deconv1
# deconv(inputs, filter_shape, strides, out_shape, is_sn, padding="SAME")
with tf.variable_scope(name_or_scope="deconv1"):
output = deconv(output, [3, 3, 256, 522], [1, 2, 2, 1], [batch_size, 7, 7, 256], padding="SAME")
output = bn(output)
output = tf.nn.relu(output)
output = conv_cond_concat(output,label_)
print("g_deconv1:", output)
# deconv2
with tf.variable_scope(name_or_scope="deconv2"):
output = deconv(output, [3, 3, 128, 266], [1, 2, 2, 1], [batch_size, 14, 14, 128], padding="SAME")
output = bn(output)
output = tf.nn.relu(output)
output = conv_cond_concat(output,label_)
print("g_deconv2:", output)
# deconv3
with tf.variable_scope(name_or_scope="deconv3"):
output = deconv(output, [5, 5, 64, 138], [1, 2, 2, 1], [batch_size, 28, 28, 64], padding="SAME")
output = bn(output)
output = tf.nn.relu(output)
output = conv_cond_concat(output,label_)
print("g_deconv5:", output)
# deconv4
with tf.variable_scope(name_or_scope="deconv4"):
output = deconv(output, [5, 5, channel, 74], [1, 1, 1, 1], [batch_size, width, height, channel], padding="SAME")
output = tf.nn.tanh(output)
print("g_deconv4:", output)
return output
@property
def var(self):
# 生成器所有变量
return tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, self.name)
class Discriminator:
def __init__(self, name):
self.name = name
def __call__(self, inputs, label, reuse=False, is_sn=False):
with tf.variable_scope(name_or_scope=self.name, reuse=reuse):
print("d_inputs:", inputs.shape)
# conv1
# conv(inputs, filter_shape, strides, is_sn, padding="SAME")
label = tf.reshape(label,[batch_size, 1, 1, n_class])
# 将标签concat到每一个通道的后面,中间隐藏层concat效果会提升
inputs = conv_cond_concat(inputs,label)
print("after_concat:", inputs.shape)
with tf.variable_scope("conv1"):
output = conv(inputs, [5, 5, 11, 64], [1, 1, 1, 1], is_sn, padding="SAME")
output = bn(output)
output = leaky_relu(output)
output = conv_cond_concat(output,label)
print("d_conv1:", output)
# conv2
with tf.variable_scope("conv2"):
output = conv(output, [3, 3, 74, 128], [1, 2, 2, 1], is_sn, padding="SAME")
output = bn(output)
output = leaky_relu(output)
output = conv_cond_concat(output,label)
print("d_conv2:", output)
# conv5
with tf.variable_scope("conv5"):
output = conv(output, [5, 5, 138, 256], [1, 2, 2, 1], is_sn, padding="SAME")
output = bn(output)
output = leaky_relu(output)
output = conv_cond_concat(output, label)
print("d_conv5:", output)
# conv3
with tf.variable_scope("conv3"):
output = conv(output, [3, 3, 266, 512], [1, 2, 2, 1], is_sn, padding="SAME")
output = bn(output)
output = leaky_relu(output)
output = conv_cond_concat(output, label)
print("d_conv3:", output)
output = tf.contrib.layers.flatten(output)
output = fully_connected(output, 1, is_sn)
print("d_fc:", output)
return output
@property
def var(self):
# 判别器所有变量
return tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope=self.name)
class GAN:
#Architecture of generator and discriminator just like DCGAN.
def __init__(self):
self.Z = tf.placeholder("float", [batch_size, 100])
self.img = tf.placeholder("float", [batch_size, width, height, channel])
self.label = tf.placeholder("float",[batch_size, n_class])
D = Discriminator("discriminator")
G = Generator("generator")
self.fake_img = G(self.Z,self.label)
if GAN_type == "DCGAN":
#DCGAN, paper: UNSUPERVISED REPRESENTATION LEARNING WITH DEEP CONVOLUTIONAL GENERATIVE ADVERSARIAL NETWORKS
self.fake_logit = tf.nn.sigmoid(D(self.fake_img,self.label))
self.real_logit = tf.nn.sigmoid(D(self.img,self.label,reuse=True))
self.d_loss = - (tf.reduce_mean(tf.log(self.real_logit + epsilon)) + tf.reduce_mean(tf.log(1 - self.fake_logit + epsilon)))
self.g_loss = - tf.reduce_mean(tf.log(self.fake_logit + epsilon))
self.opt_D = tf.train.AdamOptimizer(2e-4, beta1=0.5).minimize(self.d_loss, var_list=D.var)
self.opt_G = tf.train.AdamOptimizer(2e-4, beta1=0.5).minimize(self.g_loss, var_list=G.var)
elif GAN_type == "WGAN":
#WGAN, paper: Wasserstein GAN
self.fake_logit = D(self.fake_img,self.label)
self.real_logit = D(self.img ,self.label,reuse=True)
self.d_loss = -tf.reduce_mean(self.real_logit) + tf.reduce_mean(self.fake_logit)
self.g_loss = -tf.reduce_mean(self.fake_logit)
self.clip = []
for _, var in enumerate(D.var):
self.clip.append(tf.clip_by_value(var, -0.01, 0.01))
self.opt_D = tf.train.RMSPropOptimizer(5e-5).minimize(self.d_loss, var_list=D.var)
self.opt_G = tf.train.RMSPropOptimizer(5e-5).minimize(self.g_loss, var_list=G.var)
elif GAN_type == "WGAN-GP":
#WGAN-GP, paper: Improved Training of Wasserstein GANs
self.fake_logit = D(self.fake_img,self.label)
self.real_logit = D(self.img,self.label, reuse=True)
e = tf.random_uniform([batch_size, 1, 1, 1], 0, 1)
x_hat = e * self.img + (1 - e) * self.fake_img
grad = tf.gradients(D(x_hat, self.label,reuse=True), x_hat)[0]
self.d_loss = tf.reduce_mean(self.fake_logit - self.real_logit) + 10 * tf.reduce_mean(tf.square(tf.sqrt(tf.reduce_sum(tf.square(grad), axis=[1, 2, 3])) - 1))
self.g_loss = tf.reduce_mean(-self.fake_logit)
self.opt_D = tf.train.AdamOptimizer(1e-4, beta1=0., beta2=0.9).minimize(self.d_loss, var_list=D.var)
self.opt_G = tf.train.AdamOptimizer(1e-4, beta1=0., beta2=0.9).minimize(self.g_loss, var_list=G.var)
elif GAN_type == "LSGAN":
#LSGAN, paper: Least Squares Generative Adversarial Networks
self.fake_logit = D(self.fake_img,self.label)
self.real_logit = D(self.img, self.label,reuse=True)
self.d_loss = tf.reduce_mean(0.5 * tf.square(self.real_logit - 1) + 0.5 * tf.square(self.fake_logit))
self.g_loss = tf.reduce_mean(0.5 * tf.square(self.fake_logit - 1))
self.opt_D = tf.train.AdamOptimizer(2e-4, beta1=0.5).minimize(self.d_loss, var_list=D.var)
self.opt_G = tf.train.AdamOptimizer(2e-4, beta1=0.5).minimize(self.g_loss, var_list=G.var)
elif GAN_type == "SNGAN":
#SNGAN, paper: SPECTRAL NORMALIZATION FOR GENERATIVE ADVERSARIAL NETWORKS
self.fake_logit = tf.nn.sigmoid(D(self.fake_img,self.label, is_sn=True))
self.real_logit = tf.nn.sigmoid(D(self.img,self.label, reuse=True, is_sn=True))
self.d_loss = - (tf.reduce_mean(tf.log(self.real_logit + epsilon) + tf.log(1 - self.fake_logit + epsilon)))
self.g_loss = - tf.reduce_mean(tf.log(self.fake_logit + epsilon))
self.opt_D = tf.train.AdamOptimizer(1e-4, beta1=0.5).minimize(self.d_loss, var_list=D.var)
self.opt_G = tf.train.AdamOptimizer(1e-4, beta1=0.5).minimize(self.g_loss, var_list=G.var)
elif GAN_type == "RSGAN":
#RSGAN, paper: The relativistic discriminator: a key element missing from standard GAN
self.fake_logit = D(self.fake_img,self.label)
self.real_logit = D(self.img,self.label, reuse=True)
self.d_loss = - tf.reduce_mean(tf.log(tf.nn.sigmoid(self.real_logit - self.fake_logit) + epsilon))
self.g_loss = - tf.reduce_mean(tf.log(tf.nn.sigmoid(self.fake_logit - self.real_logit) + epsilon))
self.opt_D = tf.train.AdamOptimizer(2e-4, beta1=0.5).minimize(self.d_loss, var_list=D.var)
self.opt_G = tf.train.AdamOptimizer(2e-4, beta1=0.5).minimize(self.g_loss, var_list=G.var)
elif GAN_type == "RaSGAN":
#RaSGAN, paper: The relativistic discriminator: a key element missing from standard GAN
self.fake_logit = D(self.fake_img,self.label)
self.real_logit = D(self.img,self.label,reuse=True)
self.avg_fake_logit = tf.reduce_mean(self.fake_logit)
self.avg_real_logit = tf.reduce_mean(self.real_logit)
self.D_r_tilde = tf.nn.sigmoid(self.real_logit - self.avg_fake_logit)
self.D_f_tilde = tf.nn.sigmoid(self.fake_logit - self.avg_real_logit)
self.d_loss = - tf.reduce_mean(tf.log(self.D_r_tilde + epsilon)) - tf.reduce_mean(tf.log(1 - self.D_f_tilde + epsilon))
self.g_loss = - tf.reduce_mean(tf.log(self.D_f_tilde + epsilon)) - tf.reduce_mean(tf.log(1 - self.D_r_tilde + epsilon))
self.opt_D = tf.train.AdamOptimizer(2e-4, beta1=0.5).minimize(self.d_loss, var_list=D.var)
self.opt_G = tf.train.AdamOptimizer(2e-4, beta1=0.5).minimize(self.g_loss, var_list=G.var)
self.sess = tf.Session()
self.sess.run(tf.global_variables_initializer())
self.saver = tf.train.Saver()
# if tf.train.latest_checkpoint('class_checkpoint') is not None:
# self.saver.restore(self.sess, tf.train.latest_checkpoint('class_checkpoint'))
def __call__(self):
for e in range(epochs):
# 重新打乱数据和标签
np.random.seed(1000)
np.random.shuffle(data)
np.random.shuffle(label)
for i in range(len(data)//batch_size-1):
batch = data[i*batch_size:(i+1)*batch_size, :, :, :]
label_batch = label[i*batch_size:(i+1)*batch_size,:]
batch = batch * 2 - 1
# z = np.random.standard_normal([batch_size, 100])
z = np.random.uniform(-1, 1, (batch_size, 100)).astype(np.float32)
g_loss = self.sess.run(self.g_loss, feed_dict={self.img: batch, self.Z: z, self.label: label_batch})
d_loss = self.sess.run(self.d_loss, feed_dict={self.img: batch, self.Z: z, self.label: label_batch})
if GAN_type == "WGAN-GP":
for j in range(2):
self.sess.run(self.opt_D, feed_dict={self.img: batch, self.Z: z, self.label: label_batch})
else:
self.sess.run(self.opt_D, feed_dict={self.img: batch, self.Z: z, self.label: label_batch})
if GAN_type == "WGAN":
self.sess.run(self.clip)#WGAN weight clipping
self.sess.run(self.opt_G, feed_dict={self.img: batch, self.Z: z, self.label: label_batch})
if i % 10 == 0:
print("epoch: %d, step: [%d / %d], d_loss: %g, g_loss: %g" % (e, i, len(data)//batch_size, d_loss, g_loss))
#保存每一轮的图片
if i % 100 == 0:
# z = np.random.standard_normal([batch_size, 100])
z = np.random.uniform(-1, 1, (batch_size, 100)).astype(np.float32)
# 测试条件gan,生成数字6
c_digit = np.array([0,0,0,0,0,0,1,0,0,0])
label_batch = np.tile(c_digit,batch_size)
label_batch = label_batch.reshape(-1,10)
imgs = self.sess.run(self.fake_img, feed_dict={self.Z: z, self.label: label_batch})
imgs = (imgs + 1) / 2
sample_images(self,imgs, i)
self.saver.save(self.sess, "c_mnist_checkpoint/model_%d.ckpt" % e)
self.sess.close()
def test(self):
# self.saver.restore(self.sess, tf.train.latest_checkpoint("class_checkpoint"))
self.saver.restore(self.sess, "c_class_checkpoint\model_27.ckpt")
concat_gen = []
for i in range(200):
# z = np.random.standard_normal([batch_size, 100])
z = np.random.uniform(-1, 1, (batch_size, 100)).astype(np.float32)
G = Generator("generator")
gen = self.sess.run(G(self.Z, reuse=True), feed_dict={self.Z: z})
gen = gen.reshape(-1, 16, 16)
gen = (gen + 1) / 2
for j in range(len(gen)):
concat_gen.append(gen[j])
gen = np.array(concat_gen)
print(gen.shape)
self.sess.close()
if __name__ == "__main__":
gan = GAN()
gan()
# gan.test()
LOSS值
生成数字6的效果
https://blog.csdn.net/wspba/article/details/54730871
https://github.com/carpedm20/DCGAN-tensorflow
