def show_generator_output(sess, n_images, z_input, out_channel_dim,
image_mode):
"""
Show example output for the generator
:param sess: TensorFlow session
:param n_images: Number of Images to display
:param input_z: Input Z Tensor
:param out_channel_dim: The number of channels in the output image
:param image_mode: The mode to use for images ("RGB" or "L")
"""
cmap = None if image_mode == 'RGB' else 'gray' #设置灰度还是彩色图片
z_dim = np.shape(z_input)[-1]
example_z = np.random.uniform(-1, 1, size=(n_images, z_dim))
#print(np.shape(example_z),'example_z')
#test=generator(example_z, out_channel_dim, is_train=False,reuse=True)
result = sess.run(
generator(example_z, out_channel_dim, is_train=False,
reuse=True)) #False代表的是非训练过程，这里的参数是reuse=true
#feed_dict={input_z: example_z})
images_grid = helper.images_square_grid(result, image_mode)
pyplot.imshow(images_grid, cmap=cmap)
pyplot.show()
def output_fig(images_array, file_name="./results"):
# the shape of your images_array should be (9, width, height, 3),
# 28 <= width, height <= 112
plt.figure(figsize=(6, 6), dpi=100)
plt.imshow(helper.images_square_grid(images_array))
plt.axis("off")
plt.savefig(file_name + '.png', bbox_inches='tight', pad_inches=0)
def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode, image_path, save, show):
"""
Show example output for the generator
:param sess: TensorFlow session
:param n_images: Number of Images to display
:param input_z: Input Z Tensor
:param out_channel_dim: The number of channels in the output image
:param image_mode: The mode to use for images ("RGB" or "L")
:param image_path: Path to save the image
"""
cmap = None if image_mode == 'RGB' else 'gray'
z_dim = input_z.get_shape().as_list()[-1]
example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])
samples = sess.run(
generator(input_z, out_channel_dim, False),
feed_dict={input_z: example_z})
images_grid = helper.images_square_grid(samples, image_mode)
if save == True:
# Save image
images_grid.save(image_path, 'JPEG')
if show == True:
plt.imshow(images_grid, cmap=cmap)
plt.show()
def view_sample(gen_samples, data_image_mode):
mosaic = helper.images_square_grid(gen_samples, data_image_mode)
if data_image_mode == 'L':
pyplot.imshow(mosaic, cmap='gray')
else:
pyplot.imshow(mosaic)
pyplot.show()
def display_examples(n_images):
"""
Display an example of MNIST & CelebA
:param n_images: number of images to display
"""
show_n_images = n_images
get_ipython().magic('matplotlib inline')
mnist_images = helper.get_batch(glob(os.path.join(DATA_DIR, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
plt.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')
plt.show()
mnist_images = helper.get_batch(glob(os.path.join(DATA_DIR, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
plt.imshow(helper.images_square_grid(mnist_images, 'RGB'))
plt.show()
def output_fig(images_array, file_name="./results"):
plt.figure(figsize=(6, 6), dpi=100)
print(images_array.shape)
#plt.imshow(np.transpose(images_array.cpu().detach().numpy(),(1, 2, 0)))
plt.imshow(helper.images_square_grid(images_array))
plt.axis("off")
plt.savefig(file_name + '.png', bbox_inches='tight', pad_inches=0)
plt.close('all')
def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
cmap = None if image_mode == 'RGB' else 'gray'
z_dim = input_z.get_shape().as_list()[-1]
example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])
samples = sess.run(
generator(input_z, out_channel_dim, False),
feed_dict={input_z: example_z})
images_grid = helper.images_square_grid(samples, image_mode)
return images_grid
def show_generator_output(sess, n_images, input_z, out_channel_dim):
"""
Show example output for the generator
"""
z_dim = input_z.get_shape().as_list()[-1]
example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])
samples = sess.run(generator(input_z, out_channel_dim, False),
feed_dict={input_z: example_z})
pyplot.imshow(helper.images_square_grid(samples))
pyplot.show()
def train(self, epochs, batch_size=256):
steps = 0
noise_fix = np.random.normal(0, 1, (9, self.latent_dim))
valid = np.ones((batch_size, 1))
fake = np.zeros((batch_size, 1))
for epoch in range(epochs):
for batch_images in celeba_dataset.get_batches(batch_size):
batch_images *= 2
noise = np.random.normal(0, 1, (batch_size, self.latent_dim))
gen_imgs = self.generator.predict(noise)
out_noise = np.random.normal(0, 1, (9, self.latent_dim))
out_imgs = self.generator.predict(out_noise)
d_loss_real = self.discriminator.train_on_batch(
batch_images, valid)
d_loss_fake = self.discriminator.train_on_batch(gen_imgs, fake)
d_loss = 0.5 * np.add(d_loss_real, d_loss_fake)
if steps % 100 == 0:
fixed_img = self.generator.predict(noise_fix)
plt.imshow(helper.images_square_grid(out_imgs))
plt.show()
plt.imshow(helper.images_square_grid(fixed_img))
plt.show()
g_loss = self.combined.train_on_batch(noise, valid)
print("%d [D loss: %f, acc.: %.2f%%] [G loss: %f]" %
(steps, d_loss[0], 100 * d_loss[1], g_loss))
steps += 1
print("epoch = ", epoch)
epoch += 1
def show_generator_output(sess, n_images, input_z, out_channel_dim,
image_mode):
"""
Show example output for the generator
:param sess: TensorFlow session
:param n_images: Number of Images to display
:param input_z: Input Z Tensor
:param out_channel_dim: The number of channels in the output image
:param image_mode: The mode to use for images ("RGB" or "L")
"""
cmap = None if image_mode == 'RGB' else 'gray'
z_dim = input_z.get_shape().as_list()[-1]
example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])
samples = sess.run(generator(input_z, out_channel_dim, False),
feed_dict={input_z: example_z})
if cmap == 'gray':
pyplot.imshow(np.array(helper.images_square_grid(samples, image_mode)))
pyplot.show()
else:
images_grid = helper.images_square_grid(samples, image_mode)
pyplot.imshow(images_grid, cmap=cmap)
pyplot.show()
def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
"""
Show example output for the generator
:param sess: TensorFlow session
:param n_images: Number of Images to display
:param input_z: Input Z Tensor
:param out_channel_dim: The number of channels in the output image
:param image_mode: The mode to use for images ("RGB" or "L")
"""
cmap = None if image_mode == 'RGB' else 'gray'
z_dim = input_z.get_shape().as_list()[-1]
example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])
samples = sess.run(
generator(input_z, out_channel_dim, False),
feed_dict={input_z: example_z})
images_grid = helper.images_square_grid(samples, image_mode)
pyplot.imshow(images_grid, cmap=cmap)
pyplot.show()
def show_generator_output(sess, n_images, input_z, out_channel_dim,
image_mode):
cmap = None if image_mode == 'RGB' else 'gray'
z_dim = input_z.get_shape().as_list()[-1]
example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])
samples = sess.run(generator(input_z, out_channel_dim, False),
feed_dict={input_z: example_z})
# print(samples.shape)
images_grid = helper.images_square_grid(
samples,
image_mode) #if want to show process of tranning uncommnet this line
pyplot.imshow(
images_grid,
cmap=cmap) #if want to show process of tranning uncommnet this line
pyplot.axis(
'off') #if want to show process of tranning uncommnet this line
pyplot.show() #if want to show process of tranning uncommnet this line
return images_grid #if want to show process of tranning uncommnet this line
def output_fig(self, images_array, file_name):
plt.figure(figsize=(6, 6), dpi=100)
plt.imshow(helper.images_square_grid(images_array))
plt.axis("off")
plt.savefig(file_name + '.png', bbox_inches='tight', pad_inches=0)
# [MNIST](http://yann.lecun.com/exdb/mnist/) 是一个手写数字的图像数据集。你可以更改 `show_n_images` 探索此数据集。
# In[2]:
show_n_images = 25
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
get_ipython().run_line_magic('matplotlib', 'inline')
import os
from glob import glob
from matplotlib import pyplot
mnist_images = helper.get_batch(
glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')
# ### CelebA
# [CelebFaces Attributes Dataset (CelebA)](http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html) 是一个包含 20 多万张名人图片及相关图片说明的数据集。你将用此数据集生成人脸，不会用不到相关说明。你可以更改 `show_n_images` 探索此数据集。
# In[3]:
show_n_images = 25
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(
glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28,
28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))
show_n_images = 25
data_dir = './data'
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
get_ipython().run_line_magic('matplotlib', 'inline')
import helper
import os
from glob import glob
from matplotlib import pyplot
import numpy as np
mnist_images = helper.get_batch(
glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(np.array(helper.images_square_grid(mnist_images, 'L')))
# ### CelebA
# The [CelebFaces Attributes Dataset (CelebA)](http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing `show_n_images`.
# In[3]:
show_n_images = 25
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(
glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28,
28, 'RGB')
helper.images_square_grid(mnist_images, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))
# [MNIST](http://yann.lecun.com/exdb/mnist/) 是一个手写数字的图像数据集。你可以更改 `show_n_images` 探索此数据集。 # In[90]: show_n_images = 25 """ DON'T MODIFY ANYTHING IN THIS CELL """ get_ipython().magic(u'matplotlib inline') import os from glob import glob from matplotlib import pyplot mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L') pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray') # ### CelebA # [CelebFaces Attributes Dataset (CelebA)](http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html) 是一个包含 20 多万张名人图片及相关图片说明的数据集。你将用此数据集生成人脸，不会用不到相关说明。你可以更改 `show_n_images` 探索此数据集。 # In[91]: show_n_images = 25 """ DON'T MODIFY ANYTHING IN THIS CELL """ mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB') pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))
def output_fig(images_array, file_name):
plt.figure(figsize=(6, 6), dpi=100)
plt.imshow(helper.images_square_grid(images_array))
plt.axis("off")
plt.savefig(file_name + ".jpg", bbox_inches="tight", pad_inches=0)
plt.close()
import helper
import os
from glob import glob
import numpy as np
from matplotlib import pyplot
from distutils.version import LooseVersion
import warnings
import tensorflow as tf
import problem_unittests as tests
show_n_images = 40
celeb_images = helper.get_batch(
glob(os.path.join('img_align_celeba/*.jpg'))[:show_n_images], 50, 50,
'RGB')
pyplot.imshow(helper.images_square_grid(celeb_images, 'RGB'))
# pyplot.show()
# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion(
'1.0'
), 'Please use TensorFlow version 1.0 or newer. You are using {}'.format(
tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))
if not tf.test.gpu_device_name():
warnings.warn(
'No GPU found. Please use a GPU to train your neural network.')
else:
print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))
# Train generator
nd_noise = np.random.normal(0,1,(batch_size, noise_size))
# nd_noise = np.random.uniform(-1, 1, (batch_size, noise_size))
gac_sum += G_train.train_on_batch(nd_noise, mones)
accnt += 1
if(bcnt%4992==0):
print('Batch no.', bcnt)
print('Discriminator ave real loss:', dacr_sum / accnt / n_critic)
print('Discriminator ave fake loss:', dacf_sum / accnt / n_critic)
print('Discriminator ave loss:', (dacr_sum[0] + dacf_sum[0]) / accnt / n_critic)
print('Generator ave loss: %f %f' % tuple(gac_sum / accnt))
nd_noise = np.random.normal(0, 1, (batch_size, noise_size))
# nd_noise = np.random.uniform(-1, 1, (batch_size, noise_size))
fake_img = G.predict(nd_noise)
hp.images_square_grid(fake_img).save('outputs/' + 'iter-' + str(epch) + '-' + str(int(bcnt/9984)) + '.png')
if(bcnt%(9984*10)==0):
D.save_weights('./weights/D/D' + '-iter-' + str(epch) + '-' + str(bcnt/(9984*10)))
G.save_weights('./weights/G/G' + '-iter-' + str(epch) + '-' + str(bcnt/(9984*10)))
with open('log.txt','a') as w:
w.write('Iter-' + str(epch) + '-' + str(bcnt/(9984*10)) + '\n')
w.write(str(dacr_sum / accnt / n_critic) + '\n')
w.write(str(dacf_sum / accnt / n_critic) + '\n')
w.write(str((dacr_sum[0] + dacf_sum[0]) / accnt / n_critic) + '\n')
w.write(str(gac_sum / accnt) + '\n')
print('Model saved!')
accnt = 0
dacr_sum = np.zeros(2)
dacf_sum = np.zeros(2)
import os, time, multiprocessing
import numpy as np
import tensorflow as tf
import tensorlayer as tl
from glob import glob
from data import get_celebA, flags
from model import get_generator, get_discriminator
import helper
import matplotlib.pyplot as plt
plt.switch_backend('agg')
if __name__ == '__main__':
G = get_generator([None, flags.z_dim])
G.load_weights('checkpoint/G12.npz')
plt.figure(figsize=(6, 6), dpi=100)
for i in range(500):
z = np.random.normal(loc=0.0, scale=1.0,
size=[9, flags.z_dim]).astype(np.float32)
G.eval()
result = G(z)
G.train()
plt.imshow(helper.images_square_grid(result.numpy()))
plt.axis("off")
imagePath = '{}/{:003d}_image.png'.format(flags.sample_dir, i + 1)
if os.path.exists(imagePath):
continue
plt.savefig(imagePath, bbox_inches='tight', pad_inches=0)
glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
#pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')
# ### CelebA
# The [CelebFaces Attributes Dataset (CelebA)](http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing `show_n_images`.
# In[3]:
show_n_images = 25
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(
glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28,
28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))
# ## Preprocess the Data
# Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.
#
# The MNIST images are black and white images with a single [color channel](https://en.wikipedia.org/wiki/Channel_(digital_image%29) while the CelebA images have [3 color channels (RGB color channel)](https://en.wikipedia.org/wiki/Channel_(digital_image%29#RGB_Images).
# ## Build the Neural Network
# You'll build the components necessary to build a GANs by implementing the following functions below:
# - `model_inputs`
# - `discriminator`
# - `generator`
# - `model_loss`
# - `model_opt`
# - `train`
#
# ### Check the Version of TensorFlow and Access to GPU
IMAGE_MAX_VALUE = 255
current_index = 0
while current_index + batch_size <= shape[0]:
data_batch = get_batch(
data_files[current_index:current_index + batch_size], *shape[1:3])
current_index += batch_size
yield data_batch / IMAGE_MAX_VALUE - 0.5
#%%
test_images = get_batch(glob(os.path.join(data_dir, '*.jpg'))[:10], 32, 32)
pyplot.imshow(helper.images_square_grid(test_images))
#%%
import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()
def model_inputs(image_width, image_height, image_channels, z_dim):
"""
Create the model inputs
"""
inputs_real = tf.placeholder(tf.float32,
shape=(None, image_width, image_height,
image_channels),
name='input_real')
def show_images(ids):
show_n_images = 16
celeb_images = helper.get_batch(ids[:show_n_images], 28, 28, 'RGB')
plt.imshow(helper.images_square_grid(celeb_images, 'RGB'))
else:
print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.8)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
import keras as K
K.backend.clear_session()
# In[ ]:
if __name__ == '__main__':
bgan = DCGAN()
bgan.train(epochs=30, batch_size=128)
# In[5]:
for i in range(500):
noise = np.random.normal(0, 1, (9, 100))
plt.imshow(helper.images_square_grid(bgan.generator.predict(noise)))
plt.axis("off")
plt.savefig("./data/face_images_2/ %d_image.png" % i)
plt.show()
# In[6]:
bgan.train(epochs=10, batch_size=128)
# In[ ]:
kernel_initializer=rinit)
])
if __name__ == '__main__':
G.summary()
lo_it = 3
lo_ba = 0
pfix = '-iter-' + str(lo_it) + '-' + str(lo_ba) + '.0'
try:
G.load_weights('./weights/G/G' + pfix)
except:
print('G weights not found!')
input()
G.summary()
for i in range(500):
nd_noise = np.random.normal(0, 1, (batch_size, noise_size))
#genimg = ( * 127.5) + 127.5
gridimg = hp.images_square_grid(G.predict(nd_noise))
gridimg.save('imggen/' + ('%03d' % i) + '_image.png')
'''
img_size = 64
batch_size = 9
path = '../dataset/img_align_celeba/'
dir = np.array([path + fname for fname in os.listdir(path)])
select = np.random.randint(0, dir.shape[0], batch_size)
real_image = (hp.get_batch(dir[select], img_size, img_size, 'RGB') - 127.5) / 127.5
imgg = hp.images_square_grid(real_image)
imgg.save('grid.png')
'''
