1. LSTM Text Generation
解析:
'''
Example script to generate text from Nietzsche's writings.
'''
from __future__ import print_function
from keras.models import Sequential
from keras.layers import Dense, Activation
from keras.layers import LSTM
from keras.optimizers import RMSprop
from keras.utils.data_utils import get_file
import numpy as np
import random
import sys
path = get_file('nietzsche.txt', origin='https://s3.amazonaws.com/text-datasets/nietzsche.txt')
text = open(path).read().lower()
print('corpus length:', len(text))
chars = sorted(list(set(text)))
print('total chars:', len(chars))
char_indices = dict((c, i) for i, c in enumerate(chars))
indices_char = dict((i, c) for i, c in enumerate(chars))
# cut the text in semi-redundant sequences of maxlen characters
maxlen = 40
step = 3
sentences = []
next_chars = []
for i in range(0, len(text) - maxlen, step):
sentences.append(text[i: i + maxlen])
next_chars.append(text[i + maxlen])
print('nb sequences:', len(sentences))
print('Vectorization...')
x = np.zeros((len(sentences), maxlen, len(chars)), dtype=np.bool)
y = np.zeros((len(sentences), len(chars)), dtype=np.bool)
for i, sentence in enumerate(sentences):
for t, char in enumerate(sentence):
x[i, t, char_indices[char]] = 1
y[i, char_indices[next_chars[i]]] = 1
# build the model: a single LSTM
print('Build model...')
model = Sequential()
model.add(LSTM(128, input_shape=(maxlen, len(chars))))
model.add(Dense(len(chars)))
model.add(Activation('softmax'))
optimizer = RMSprop(lr=0.01)
model.compile(loss='categorical_crossentropy', optimizer=optimizer)
def sample(preds, temperature=1.0):
# helper function to sample an index from a probability array
preds = np.asarray(preds).astype('float64')
preds = np.log(preds) / temperature
exp_preds = np.exp(preds)
preds = exp_preds / np.sum(exp_preds)
probas = np.random.multinomial(1, preds, 1)
return np.argmax(probas)
# train the model, output generated text after each iteration
for iteration in range(1, 60):
print()
print('-' * 50)
print('Iteration', iteration)
model.fit(x, y,
batch_size=128,
epochs=1)
start_index = random.randint(0, len(text) - maxlen - 1)
for diversity in [0.2, 0.5, 1.0, 1.2]:
print()
print('----- diversity:', diversity)
generated = ''
sentence = text[start_index: start_index + maxlen]
generated += sentence
print('----- Generating with seed: "' + sentence + '"')
sys.stdout.write(generated)
for i in range(400):
x_pred = np.zeros((1, maxlen, len(chars)))
for t, char in enumerate(sentence):
x_pred[0, t, char_indices[char]] = 1.
preds = model.predict(x_pred, verbose=0)[0]
next_index = sample(preds, diversity)
next_char = indices_char[next_index]
generated += next_char
sentence = sentence[1:] + next_char
sys.stdout.write(next_char)
sys.stdout.flush()
print()
2. Conv Filter Visualization
解析:
'''
Visualization of the filters of VGG16, via gradient ascent in input space.
'''
from __future__ import print_function
from scipy.misc import imsave
import numpy as np
import time
from keras.applications import vgg16
from keras import backend as K
# dimensions of the generated pictures for each filter.
img_width = 128
img_height = 128
# the name of the layer we want to visualize
# (see model definition at keras/applications/vgg16.py)
layer_name = 'block5_conv1'
# util function to convert a tensor into a valid image
def deprocess_image(x):
# normalize tensor: center on 0., ensure std is 0.1
x -= x.mean()
x /= (x.std() + 1e-5)
x *= 0.1
# clip to [0, 1]
x += 0.5
x = np.clip(x, 0, 1)
# convert to RGB array
x *= 255
if K.image_data_format() == 'channels_first':
x = x.transpose((1, 2, 0))
x = np.clip(x, 0, 255).astype('uint8')
return x
# build the VGG16 network with ImageNet weights
model = vgg16.VGG16(weights='imagenet', include_top=False)
print('Model loaded.')
model.summary()
# this is the placeholder for the input images
input_img = model.input
# get the symbolic outputs of each "key" layer (we gave them unique names).
layer_dict = dict([(layer.name, layer) for layer in model.layers[1:]])
def normalize(x):
# utility function to normalize a tensor by its L2 norm
return x / (K.sqrt(K.mean(K.square(x))) + 1e-5)
kept_filters = []
for filter_index in range(200):
# we only scan through the first 200 filters,
# but there are actually 512 of them
print('Processing filter %d' % filter_index)
start_time = time.time()
# we build a loss function that maximizes the activation
# of the nth filter of the layer considered
layer_output = layer_dict[layer_name].output
if K.image_data_format() == 'channels_first':
loss = K.mean(layer_output[:, filter_index, :, :])
else:
loss = K.mean(layer_output[:, :, :, filter_index])
# we compute the gradient of the input picture wrt this loss
grads = K.gradients(loss, input_img)[0]
# normalization trick: we normalize the gradient
grads = normalize(grads)
# this function returns the loss and grads given the input picture
iterate = K.function([input_img], [loss, grads])
# step size for gradient ascent
step = 1.
# we start from a gray image with some random noise
if K.image_data_format() == 'channels_first':
input_img_data = np.random.random((1, 3, img_width, img_height))
else:
input_img_data = np.random.random((1, img_width, img_height, 3))
input_img_data = (input_img_data - 0.5) * 20 + 128
# we run gradient ascent for 20 steps
for i in range(20):
loss_value, grads_value = iterate([input_img_data])
input_img_data += grads_value * step
print('Current loss value:', loss_value)
if loss_value <= 0.:
# some filters get stuck to 0, we can skip them
break
# decode the resulting input image
if loss_value > 0:
img = deprocess_image(input_img_data[0])
kept_filters.append((img, loss_value))
end_time = time.time()
print('Filter %d processed in %ds' % (filter_index, end_time - start_time))
# we will stich the best 64 filters on a 8 x 8 grid.
n = 8
# the filters that have the highest loss are assumed to be better-looking.
# we will only keep the top 64 filters.
kept_filters.sort(key=lambda x: x[1], reverse=True)
kept_filters = kept_filters[:n * n]
# build a black picture with enough space for
# our 8 x 8 filters of size 128 x 128, with a 5px margin in between
margin = 5
width = n * img_width + (n - 1) * margin
height = n * img_height + (n - 1) * margin
stitched_filters = np.zeros((width, height, 3))
# fill the picture with our saved filters
for i in range(n):
for j in range(n):
img, loss = kept_filters[i * n + j]
stitched_filters[(img_width + margin) * i: (img_width + margin) * i + img_width,
(img_height + margin) * j: (img_height + margin) * j + img_height, :] = img
# save the result to disk
imsave('stitched_filters_%dx%d.png' % (n, n), stitched_filters)
3. Deep Dream
解析:
'''Deep Dreaming in Keras.
Run the script with:
```
python deep_dream.py path_to_your_base_image.jpg prefix_for_results
```
e.g.:
```
python deep_dream.py img/mypic.jpg results/dream
```
'''
from __future__ import print_function
from keras.preprocessing.image import load_img, img_to_array
import numpy as np
import scipy
import argparse
from keras.applications import inception_v3
from keras import backend as K
parser = argparse.ArgumentParser(description='Deep Dreams with Keras.')
parser.add_argument('base_image_path', metavar='base', type=str,
help='Path to the image to transform.')
parser.add_argument('result_prefix', metavar='res_prefix', type=str,
help='Prefix for the saved results.')
args = parser.parse_args()
base_image_path = args.base_image_path
result_prefix = args.result_prefix
# These are the names of the layers
# for which we try to maximize activation,
# as well as their weight in the final loss
# we try to maximize.
# You can tweak these setting to obtain new visual effects.
settings = {
'features': {
'mixed2': 0.2,
'mixed3': 0.5,
'mixed4': 2.,
'mixed5': 1.5,
},
}
def preprocess_image(image_path):
# Util function to open, resize and format pictures
# into appropriate tensors.
img = load_img(image_path)
img = img_to_array(img)
img = np.expand_dims(img, axis=0)
img = inception_v3.preprocess_input(img)
return img
def deprocess_image(x):
# Util function to convert a tensor into a valid image.
if K.image_data_format() == 'channels_first':
x = x.reshape((3, x.shape[2], x.shape[3]))
x = x.transpose((1, 2, 0))
else:
x = x.reshape((x.shape[1], x.shape[2], 3))
x /= 2.
x += 0.5
x *= 255.
x = np.clip(x, 0, 255).astype('uint8')
return x
K.set_learning_phase(0)
# Build the InceptionV3 network with our placeholder.
# The model will be loaded with pre-trained ImageNet weights.
model = inception_v3.InceptionV3(weights='imagenet',
include_top=False)
dream = model.input
print('Model loaded.')
# Get the symbolic outputs of each "key" layer (we gave them unique names).
layer_dict = dict([(layer.name, layer) for layer in model.layers])
# Define the loss.
loss = K.variable(0.)
for layer_name in settings['features']:
# Add the L2 norm of the features of a layer to the loss.
assert layer_name in layer_dict.keys(), 'Layer ' + layer_name + ' not found in model.'
coeff = settings['features'][layer_name]
x = layer_dict[layer_name].output
# We avoid border artifacts by only involving non-border pixels in the loss.
scaling = K.prod(K.cast(K.shape(x), 'float32'))
if K.image_data_format() == 'channels_first':
loss += coeff * K.sum(K.square(x[:, :, 2: -2, 2: -2])) / scaling
else:
loss += coeff * K.sum(K.square(x[:, 2: -2, 2: -2, :])) / scaling
# Compute the gradients of the dream wrt the loss.
grads = K.gradients(loss, dream)[0]
# Normalize gradients.
grads /= K.maximum(K.mean(K.abs(grads)), 1e-7)
# Set up function to retrieve the value
# of the loss and gradients given an input image.
outputs = [loss, grads]
fetch_loss_and_grads = K.function([dream], outputs)
def eval_loss_and_grads(x):
outs = fetch_loss_and_grads([x])
loss_value = outs[0]
grad_values = outs[1]
return loss_value, grad_values
def resize_img(img, size):
img = np.copy(img)
if K.image_data_format() == 'channels_first':
factors = (1, 1,
float(size[0]) / img.shape[2],
float(size[1]) / img.shape[3])
else:
factors = (1,
float(size[0]) / img.shape[1],
float(size[1]) / img.shape[2],
1)
return scipy.ndimage.zoom(img, factors, order=1)
def gradient_ascent(x, iterations, step, max_loss=None):
for i in range(iterations):
loss_value, grad_values = eval_loss_and_grads(x)
if max_loss is not None and loss_value > max_loss:
break
print('..Loss value at', i, ':', loss_value)
x += step * grad_values
return x
def save_img(img, fname):
pil_img = deprocess_image(np.copy(img))
scipy.misc.imsave(fname, pil_img)
"""Process:
- Load the original image.
- Define a number of processing scales (i.e. image shapes),
from smallest to largest.
- Resize the original image to the smallest scale.
- For every scale, starting with the smallest (i.e. current one):
- Run gradient ascent
- Upscale image to the next scale
- Reinject the detail that was lost at upscaling time
- Stop when we are back to the original size.
To obtain the detail lost during upscaling, we simply
take the original image, shrink it down, upscale it,
and compare the result to the (resized) original image.
"""
# Playing with these hyperparameters will also allow you to achieve new effects
step = 0.01 # Gradient ascent step size
num_octave = 3 # Number of scales at which to run gradient ascent
octave_scale = 1.4 # Size ratio between scales
iterations = 20 # Number of ascent steps per scale
max_loss = 10.
img = preprocess_image(base_image_path)
if K.image_data_format() == 'channels_first':
original_shape = img.shape[2:]
else:
original_shape = img.shape[1:3]
successive_shapes = [original_shape]
for i in range(1, num_octave):
shape = tuple([int(dim / (octave_scale ** i)) for dim in original_shape])
successive_shapes.append(shape)
successive_shapes = successive_shapes[::-1]
original_img = np.copy(img)
shrunk_original_img = resize_img(img, successive_shapes[0])
for shape in successive_shapes:
print('Processing image shape', shape)
img = resize_img(img, shape)
img = gradient_ascent(img,
iterations=iterations,
step=step,
max_loss=max_loss)
upscaled_shrunk_original_img = resize_img(shrunk_original_img, shape)
same_size_original = resize_img(original_img, shape)
lost_detail = same_size_original - upscaled_shrunk_original_img
img += lost_detail
shrunk_original_img = resize_img(original_img, shape)
save_img(img, fname=result_prefix + '.png')
4. Neural Doodle
解析:
'''
Neural doodle with Keras
'''
from __future__ import print_function
import time
import argparse
import numpy as np
from scipy.optimize import fmin_l_bfgs_b
from scipy.misc import imread, imsave
from keras import backend as K
from keras.layers import Input, AveragePooling2D
from keras.models import Model
from keras.preprocessing.image import load_img, img_to_array
from keras.applications import vgg19
# Command line arguments
parser = argparse.ArgumentParser(description='Keras neural doodle example')
parser.add_argument('--nlabels', type=int,
help='number of semantic labels'
' (regions in differnet colors)'
' in style_mask/target_mask')
parser.add_argument('--style-image', type=str,
help='path to image to learn style from')
parser.add_argument('--style-mask', type=str,
help='path to semantic mask of style image')
parser.add_argument('--target-mask', type=str,
help='path to semantic mask of target image')
parser.add_argument('--content-image', type=str, default=None,
help='path to optional content image')
parser.add_argument('--target-image-prefix', type=str,
help='path prefix for generated results')
args = parser.parse_args()
style_img_path = args.style_image
style_mask_path = args.style_mask
target_mask_path = args.target_mask
content_img_path = args.content_image
target_img_prefix = args.target_image_prefix
use_content_img = content_img_path is not None
num_labels = args.nlabels
num_colors = 3 # RGB
# determine image sizes based on target_mask
ref_img = imread(target_mask_path)
img_nrows, img_ncols = ref_img.shape[:2]
total_variation_weight = 50.
style_weight = 1.
content_weight = 0.1 if use_content_img else 0
content_feature_layers = ['block5_conv2']
# To get better generation qualities, use more conv layers for style features
style_feature_layers = ['block1_conv1', 'block2_conv1', 'block3_conv1',
'block4_conv1', 'block5_conv1']
# helper functions for reading/processing images
def preprocess_image(image_path):
img = load_img(image_path, target_size=(img_nrows, img_ncols))
img = img_to_array(img)
img = np.expand_dims(img, axis=0)
img = vgg19.preprocess_input(img)
return img
def deprocess_image(x):
if K.image_data_format() == 'channels_first':
x = x.reshape((3, img_nrows, img_ncols))
x = x.transpose((1, 2, 0))
else:
x = x.reshape((img_nrows, img_ncols, 3))
# Remove zero-center by mean pixel
x[:, :, 0] += 103.939
x[:, :, 1] += 116.779
x[:, :, 2] += 123.68
# 'BGR'->'RGB'
x = x[:, :, ::-1]
x = np.clip(x, 0, 255).astype('uint8')
return x
def kmeans(xs, k):
assert xs.ndim == 2
try:
from sklearn.cluster import k_means
_, labels, _ = k_means(xs.astype('float64'), k)
except ImportError:
from scipy.cluster.vq import kmeans2
_, labels = kmeans2(xs, k, missing='raise')
return labels
def load_mask_labels():
'''Load both target and style masks.
A mask image (nr x nc) with m labels/colors will be loaded
as a 4D boolean tensor: (1, m, nr, nc) for 'channels_first' or (1, nr, nc, m) for 'channels_last'
'''
target_mask_img = load_img(target_mask_path,
target_size=(img_nrows, img_ncols))
target_mask_img = img_to_array(target_mask_img)
style_mask_img = load_img(style_mask_path,
target_size=(img_nrows, img_ncols))
style_mask_img = img_to_array(style_mask_img)
if K.image_data_format() == 'channels_first':
mask_vecs = np.vstack([style_mask_img.reshape((3, -1)).T,
target_mask_img.reshape((3, -1)).T])
else:
mask_vecs = np.vstack([style_mask_img.reshape((-1, 3)),
target_mask_img.reshape((-1, 3))])
labels = kmeans(mask_vecs, num_labels)
style_mask_label = labels[:img_nrows *
img_ncols].reshape((img_nrows, img_ncols))
target_mask_label = labels[img_nrows *
img_ncols:].reshape((img_nrows, img_ncols))
stack_axis = 0 if K.image_data_format() == 'channels_first' else -1
style_mask = np.stack([style_mask_label == r for r in range(num_labels)],
axis=stack_axis)
target_mask = np.stack([target_mask_label == r for r in range(num_labels)],
axis=stack_axis)
return (np.expand_dims(style_mask, axis=0),
np.expand_dims(target_mask, axis=0))
# Create tensor variables for images
if K.image_data_format() == 'channels_first':
shape = (1, num_colors, img_nrows, img_ncols)
else:
shape = (1, img_nrows, img_ncols, num_colors)
style_image = K.variable(preprocess_image(style_img_path))
target_image = K.placeholder(shape=shape)
if use_content_img:
content_image = K.variable(preprocess_image(content_img_path))
else:
content_image = K.zeros(shape=shape)
images = K.concatenate([style_image, target_image, content_image], axis=0)
# Create tensor variables for masks
raw_style_mask, raw_target_mask = load_mask_labels()
style_mask = K.variable(raw_style_mask.astype('float32'))
target_mask = K.variable(raw_target_mask.astype('float32'))
masks = K.concatenate([style_mask, target_mask], axis=0)
# index constants for images and tasks variables
STYLE, TARGET, CONTENT = 0, 1, 2
# Build image model, mask model and use layer outputs as features
# image model as VGG19
image_model = vgg19.VGG19(include_top=False, input_tensor=images)
# mask model as a series of pooling
mask_input = Input(tensor=masks, shape=(None, None, None), name='mask_input')
x = mask_input
for layer in image_model.layers[1:]:
name = 'mask_%s' % layer.name
if 'conv' in layer.name:
x = AveragePooling2D((3, 3), padding='same', strides=(
1, 1), name=name)(x)
elif 'pool' in layer.name:
x = AveragePooling2D((2, 2), name=name)(x)
mask_model = Model(mask_input, x)
# Collect features from image_model and task_model
image_features = {}
mask_features = {}
for img_layer, mask_layer in zip(image_model.layers, mask_model.layers):
if 'conv' in img_layer.name:
assert 'mask_' + img_layer.name == mask_layer.name
layer_name = img_layer.name
img_feat, mask_feat = img_layer.output, mask_layer.output
image_features[layer_name] = img_feat
mask_features[layer_name] = mask_feat
# Define loss functions
def gram_matrix(x):
assert K.ndim(x) == 3
features = K.batch_flatten(x)
gram = K.dot(features, K.transpose(features))
return gram
def region_style_loss(style_image, target_image, style_mask, target_mask):
'''Calculate style loss between style_image and target_image,
for one common region specified by their (boolean) masks
'''
assert 3 == K.ndim(style_image) == K.ndim(target_image)
assert 2 == K.ndim(style_mask) == K.ndim(target_mask)
if K.image_data_format() == 'channels_first':
masked_style = style_image * style_mask
masked_target = target_image * target_mask
num_channels = K.shape(style_image)[0]
else:
masked_style = K.permute_dimensions(
style_image, (2, 0, 1)) * style_mask
masked_target = K.permute_dimensions(
target_image, (2, 0, 1)) * target_mask
num_channels = K.shape(style_image)[-1]
num_channels = K.cast(num_channels, dtype='float32')
s = gram_matrix(masked_style) / K.mean(style_mask) / num_channels
c = gram_matrix(masked_target) / K.mean(target_mask) / num_channels
return K.mean(K.square(s - c))
def style_loss(style_image, target_image, style_masks, target_masks):
'''Calculate style loss between style_image and target_image,
in all regions.
'''
assert 3 == K.ndim(style_image) == K.ndim(target_image)
assert 3 == K.ndim(style_masks) == K.ndim(target_masks)
loss = K.variable(0)
for i in range(num_labels):
if K.image_data_format() == 'channels_first':
style_mask = style_masks[i, :, :]
target_mask = target_masks[i, :, :]
else:
style_mask = style_masks[:, :, i]
target_mask = target_masks[:, :, i]
loss += region_style_loss(style_image,
target_image, style_mask, target_mask)
return loss
def content_loss(content_image, target_image):
return K.sum(K.square(target_image - content_image))
def total_variation_loss(x):
assert 4 == K.ndim(x)
if K.image_data_format() == 'channels_first':
a = K.square(x[:, :, :img_nrows - 1, :img_ncols - 1] -
x[:, :, 1:, :img_ncols - 1])
b = K.square(x[:, :, :img_nrows - 1, :img_ncols - 1] -
x[:, :, :img_nrows - 1, 1:])
else:
a = K.square(x[:, :img_nrows - 1, :img_ncols - 1, :] -
x[:, 1:, :img_ncols - 1, :])
b = K.square(x[:, :img_nrows - 1, :img_ncols - 1, :] -
x[:, :img_nrows - 1, 1:, :])
return K.sum(K.pow(a + b, 1.25))
# Overall loss is the weighted sum of content_loss, style_loss and tv_loss
# Each individual loss uses features from image/mask models.
loss = K.variable(0)
for layer in content_feature_layers:
content_feat = image_features[layer][CONTENT, :, :, :]
target_feat = image_features[layer][TARGET, :, :, :]
loss += content_weight * content_loss(content_feat, target_feat)
for layer in style_feature_layers:
style_feat = image_features[layer][STYLE, :, :, :]
target_feat = image_features[layer][TARGET, :, :, :]
style_masks = mask_features[layer][STYLE, :, :, :]
target_masks = mask_features[layer][TARGET, :, :, :]
sl = style_loss(style_feat, target_feat, style_masks, target_masks)
loss += (style_weight / len(style_feature_layers)) * sl
loss += total_variation_weight * total_variation_loss(target_image)
loss_grads = K.gradients(loss, target_image)
# Evaluator class for computing efficiency
outputs = [loss]
if isinstance(loss_grads, (list, tuple)):
outputs += loss_grads
else:
outputs.append(loss_grads)
f_outputs = K.function([target_image], outputs)
def eval_loss_and_grads(x):
if K.image_data_format() == 'channels_first':
x = x.reshape((1, 3, img_nrows, img_ncols))
else:
x = x.reshape((1, img_nrows, img_ncols, 3))
outs = f_outputs([x])
loss_value = outs[0]
if len(outs[1:]) == 1:
grad_values = outs[1].flatten().astype('float64')
else:
grad_values = np.array(outs[1:]).flatten().astype('float64')
return loss_value, grad_values
class Evaluator(object):
def __init__(self):
self.loss_value = None
self.grads_values = None
def loss(self, x):
assert self.loss_value is None
loss_value, grad_values = eval_loss_and_grads(x)
self.loss_value = loss_value
self.grad_values = grad_values
return self.loss_value
def grads(self, x):
assert self.loss_value is not None
grad_values = np.copy(self.grad_values)
self.loss_value = None
self.grad_values = None
return grad_values
evaluator = Evaluator()
# Generate images by iterative optimization
if K.image_data_format() == 'channels_first':
x = np.random.uniform(0, 255, (1, 3, img_nrows, img_ncols)) - 128.
else:
x = np.random.uniform(0, 255, (1, img_nrows, img_ncols, 3)) - 128.
for i in range(50):
print('Start of iteration', i)
start_time = time.time()
x, min_val, info = fmin_l_bfgs_b(evaluator.loss, x.flatten(),
fprime=evaluator.grads, maxfun=20)
print('Current loss value:', min_val)
# save current generated image
img = deprocess_image(x.copy())
fname = target_img_prefix + '_at_iteration_%d.png' % i
imsave(fname, img)
end_time = time.time()
print('Image saved as', fname)
print('Iteration %d completed in %ds' % (i, end_time - start_time))
5. Neural Style Transfer
解析:
'''Neural style transfer with Keras.
Run the script with:
```
python neural_style_transfer.py path_to_your_base_image.jpg path_to_your_reference.jpg prefix_for_results
```
e.g.:
```
python neural_style_transfer.py img/tuebingen.jpg img/starry_night.jpg results/my_result
```
'''
from __future__ import print_function
from keras.preprocessing.image import load_img, img_to_array
from scipy.misc import imsave
import numpy as np
from scipy.optimize import fmin_l_bfgs_b
import time
import argparse
from keras.applications import vgg19
from keras import backend as K
parser = argparse.ArgumentParser(description='Neural style transfer with Keras.')
parser.add_argument('base_image_path', metavar='base', type=str,
help='Path to the image to transform.')
parser.add_argument('style_reference_image_path', metavar='ref', type=str,
help='Path to the style reference image.')
parser.add_argument('result_prefix', metavar='res_prefix', type=str,
help='Prefix for the saved results.')
parser.add_argument('--iter', type=int, default=10, required=False,
help='Number of iterations to run.')
parser.add_argument('--content_weight', type=float, default=0.025, required=False,
help='Content weight.')
parser.add_argument('--style_weight', type=float, default=1.0, required=False,
help='Style weight.')
parser.add_argument('--tv_weight', type=float, default=1.0, required=False,
help='Total Variation weight.')
args = parser.parse_args()
base_image_path = args.base_image_path
style_reference_image_path = args.style_reference_image_path
result_prefix = args.result_prefix
iterations = args.iter
# these are the weights of the different loss components
total_variation_weight = args.tv_weight
style_weight = args.style_weight
content_weight = args.content_weight
# dimensions of the generated picture.
width, height = load_img(base_image_path).size
img_nrows = 400
img_ncols = int(width * img_nrows / height)
# util function to open, resize and format pictures into appropriate tensors
def preprocess_image(image_path):
img = load_img(image_path, target_size=(img_nrows, img_ncols))
img = img_to_array(img)
img = np.expand_dims(img, axis=0)
img = vgg19.preprocess_input(img)
return img
# util function to convert a tensor into a valid image
def deprocess_image(x):
if K.image_data_format() == 'channels_first':
x = x.reshape((3, img_nrows, img_ncols))
x = x.transpose((1, 2, 0))
else:
x = x.reshape((img_nrows, img_ncols, 3))
# Remove zero-center by mean pixel
x[:, :, 0] += 103.939
x[:, :, 1] += 116.779
x[:, :, 2] += 123.68
# 'BGR'->'RGB'
x = x[:, :, ::-1]
x = np.clip(x, 0, 255).astype('uint8')
return x
# get tensor representations of our images
base_image = K.variable(preprocess_image(base_image_path))
style_reference_image = K.variable(preprocess_image(style_reference_image_path))
# this will contain our generated image
if K.image_data_format() == 'channels_first':
combination_image = K.placeholder((1, 3, img_nrows, img_ncols))
else:
combination_image = K.placeholder((1, img_nrows, img_ncols, 3))
# combine the 3 images into a single Keras tensor
input_tensor = K.concatenate([base_image,
style_reference_image,
combination_image], axis=0)
# build the VGG16 network with our 3 images as input
# the model will be loaded with pre-trained ImageNet weights
model = vgg19.VGG19(input_tensor=input_tensor,
weights='imagenet', include_top=False)
print('Model loaded.')
# get the symbolic outputs of each "key" layer (we gave them unique names).
outputs_dict = dict([(layer.name, layer.output) for layer in model.layers])
# compute the neural style loss
# first we need to define 4 util functions
# the gram matrix of an image tensor (feature-wise outer product)
def gram_matrix(x):
assert K.ndim(x) == 3
if K.image_data_format() == 'channels_first':
features = K.batch_flatten(x)
else:
features = K.batch_flatten(K.permute_dimensions(x, (2, 0, 1)))
gram = K.dot(features, K.transpose(features))
return gram
# the "style loss" is designed to maintain
# the style of the reference image in the generated image.
# It is based on the gram matrices (which capture style) of
# feature maps from the style reference image
# and from the generated image
def style_loss(style, combination):
assert K.ndim(style) == 3
assert K.ndim(combination) == 3
S = gram_matrix(style)
C = gram_matrix(combination)
channels = 3
size = img_nrows * img_ncols
return K.sum(K.square(S - C)) / (4. * (channels ** 2) * (size ** 2))
# an auxiliary loss function
# designed to maintain the "content" of the
# base image in the generated image
def content_loss(base, combination):
return K.sum(K.square(combination - base))
# the 3rd loss function, total variation loss,
# designed to keep the generated image locally coherent
def total_variation_loss(x):
assert K.ndim(x) == 4
if K.image_data_format() == 'channels_first':
a = K.square(x[:, :, :img_nrows - 1, :img_ncols - 1] - x[:, :, 1:, :img_ncols - 1])
b = K.square(x[:, :, :img_nrows - 1, :img_ncols - 1] - x[:, :, :img_nrows - 1, 1:])
else:
a = K.square(x[:, :img_nrows - 1, :img_ncols - 1, :] - x[:, 1:, :img_ncols - 1, :])
b = K.square(x[:, :img_nrows - 1, :img_ncols - 1, :] - x[:, :img_nrows - 1, 1:, :])
return K.sum(K.pow(a + b, 1.25))
# combine these loss functions into a single scalar
loss = K.variable(0.)
layer_features = outputs_dict['block5_conv2']
base_image_features = layer_features[0, :, :, :]
combination_features = layer_features[2, :, :, :]
loss += content_weight * content_loss(base_image_features,
combination_features)
feature_layers = ['block1_conv1', 'block2_conv1',
'block3_conv1', 'block4_conv1',
'block5_conv1']
for layer_name in feature_layers:
layer_features = outputs_dict[layer_name]
style_reference_features = layer_features[1, :, :, :]
combination_features = layer_features[2, :, :, :]
sl = style_loss(style_reference_features, combination_features)
loss += (style_weight / len(feature_layers)) * sl
loss += total_variation_weight * total_variation_loss(combination_image)
# get the gradients of the generated image wrt the loss
grads = K.gradients(loss, combination_image)
outputs = [loss]
if isinstance(grads, (list, tuple)):
outputs += grads
else:
outputs.append(grads)
f_outputs = K.function([combination_image], outputs)
def eval_loss_and_grads(x):
if K.image_data_format() == 'channels_first':
x = x.reshape((1, 3, img_nrows, img_ncols))
else:
x = x.reshape((1, img_nrows, img_ncols, 3))
outs = f_outputs([x])
loss_value = outs[0]
if len(outs[1:]) == 1:
grad_values = outs[1].flatten().astype('float64')
else:
grad_values = np.array(outs[1:]).flatten().astype('float64')
return loss_value, grad_values
# this Evaluator class makes it possible
# to compute loss and gradients in one pass
# while retrieving them via two separate functions,
# "loss" and "grads". This is done because scipy.optimize
# requires separate functions for loss and gradients,
# but computing them separately would be inefficient.
class Evaluator(object):
def __init__(self):
self.loss_value = None
self.grads_values = None
def loss(self, x):
assert self.loss_value is None
loss_value, grad_values = eval_loss_and_grads(x)
self.loss_value = loss_value
self.grad_values = grad_values
return self.loss_value
def grads(self, x):
assert self.loss_value is not None
grad_values = np.copy(self.grad_values)
self.loss_value = None
self.grad_values = None
return grad_values
evaluator = Evaluator()
# run scipy-based optimization (L-BFGS) over the pixels of the generated image
# so as to minimize the neural style loss
x = preprocess_image(base_image_path)
for i in range(iterations):
print('Start of iteration', i)
start_time = time.time()
x, min_val, info = fmin_l_bfgs_b(evaluator.loss, x.flatten(),
fprime=evaluator.grads, maxfun=20)
print('Current loss value:', min_val)
# save current generated image
img = deprocess_image(x.copy())
fname = result_prefix + '_at_iteration_%d.png' % i
imsave(fname, img)
end_time = time.time()
print('Image saved as', fname)
print('Iteration %d completed in %ds' % (i, end_time - start_time))
6. Variational AutoEncoder
解析:
'''
This script demonstrates how to build a variational autoencoder with Keras.
'''
import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import norm
from keras.layers import Input, Dense, Lambda, Layer
from keras.models import Model
from keras import backend as K
from keras import metrics
from keras.datasets import mnist
batch_size = 100
original_dim = 784
latent_dim = 2
intermediate_dim = 256
epochs = 50
epsilon_std = 1.0
x = Input(shape=(original_dim,))
h = Dense(intermediate_dim, activation='relu')(x)
z_mean = Dense(latent_dim)(h)
z_log_var = Dense(latent_dim)(h)
def sampling(args):
z_mean, z_log_var = args
epsilon = K.random_normal(shape=(K.shape(z_mean)[0], latent_dim), mean=0.,
stddev=epsilon_std)
return z_mean + K.exp(z_log_var / 2) * epsilon
# note that "output_shape" isn't necessary with the TensorFlow backend
z = Lambda(sampling, output_shape=(latent_dim,))([z_mean, z_log_var])
# we instantiate these layers separately so as to reuse them later
decoder_h = Dense(intermediate_dim, activation='relu')
decoder_mean = Dense(original_dim, activation='sigmoid')
h_decoded = decoder_h(z)
x_decoded_mean = decoder_mean(h_decoded)
# Custom loss layer
class CustomVariationalLayer(Layer):
def __init__(self, **kwargs):
self.is_placeholder = True
super(CustomVariationalLayer, self).__init__(**kwargs)
def vae_loss(self, x, x_decoded_mean):
xent_loss = original_dim * metrics.binary_crossentropy(x, x_decoded_mean)
kl_loss = - 0.5 * K.sum(1 + z_log_var - K.square(z_mean) - K.exp(z_log_var), axis=-1)
return K.mean(xent_loss + kl_loss)
def call(self, inputs):
x = inputs[0]
x_decoded_mean = inputs[1]
loss = self.vae_loss(x, x_decoded_mean)
self.add_loss(loss, inputs=inputs)
# We won't actually use the output.
return x
y = CustomVariationalLayer()([x, x_decoded_mean])
vae = Model(x, y)
vae.compile(optimizer='rmsprop', loss=None)
# train the VAE on MNIST digits
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.astype('float32') / 255.
x_test = x_test.astype('float32') / 255.
x_train = x_train.reshape((len(x_train), np.prod(x_train.shape[1:])))
x_test = x_test.reshape((len(x_test), np.prod(x_test.shape[1:])))
vae.fit(x_train,
shuffle=True,
epochs=epochs,
batch_size=batch_size,
validation_data=(x_test, None))
# build a model to project inputs on the latent space
encoder = Model(x, z_mean)
# display a 2D plot of the digit classes in the latent space
x_test_encoded = encoder.predict(x_test, batch_size=batch_size)
plt.figure(figsize=(6, 6))
plt.scatter(x_test_encoded[:, 0], x_test_encoded[:, 1], c=y_test)
plt.colorbar()
plt.show()
# build a digit generator that can sample from the learned distribution
decoder_input = Input(shape=(latent_dim,))
_h_decoded = decoder_h(decoder_input)
_x_decoded_mean = decoder_mean(_h_decoded)
generator = Model(decoder_input, _x_decoded_mean)
# display a 2D manifold of the digits
n = 15 # figure with 15x15 digits
digit_size = 28
figure = np.zeros((digit_size * n, digit_size * n))
# linearly spaced coordinates on the unit square were transformed through the inverse CDF (ppf) of the Gaussian
# to produce values of the latent variables z, since the prior of the latent space is Gaussian
grid_x = norm.ppf(np.linspace(0.05, 0.95, n))
grid_y = norm.ppf(np.linspace(0.05, 0.95, n))
for i, yi in enumerate(grid_x):
for j, xi in enumerate(grid_y):
z_sample = np.array([[xi, yi]])
x_decoded = generator.predict(z_sample)
digit = x_decoded[0].reshape(digit_size, digit_size)
figure[i * digit_size: (i + 1) * digit_size,
j * digit_size: (j + 1) * digit_size] = digit
plt.figure(figsize=(10, 10))
plt.imshow(figure, cmap='Greys_r')
plt.show()
7. Variational Autoencoder Deconv
解析:
'''
This script demonstrates how to build a variational autoencoder
with Keras and deconvolution layers.
'''
import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import norm
from keras.layers import Input, Dense, Lambda, Flatten, Reshape, Layer
from keras.layers import Conv2D, Conv2DTranspose
from keras.models import Model
from keras import backend as K
from keras import metrics
from keras.datasets import mnist
# input image dimensions
img_rows, img_cols, img_chns = 28, 28, 1
# number of convolutional filters to use
filters = 64
# convolution kernel size
num_conv = 3
batch_size = 100
if K.image_data_format() == 'channels_first':
original_img_size = (img_chns, img_rows, img_cols)
else:
original_img_size = (img_rows, img_cols, img_chns)
latent_dim = 2
intermediate_dim = 128
epsilon_std = 1.0
epochs = 5
x = Input(shape=original_img_size)
conv_1 = Conv2D(img_chns,
kernel_size=(2, 2),
padding='same', activation='relu')(x)
conv_2 = Conv2D(filters,
kernel_size=(2, 2),
padding='same', activation='relu',
strides=(2, 2))(conv_1)
conv_3 = Conv2D(filters,
kernel_size=num_conv,
padding='same', activation='relu',
strides=1)(conv_2)
conv_4 = Conv2D(filters,
kernel_size=num_conv,
padding='same', activation='relu',
strides=1)(conv_3)
flat = Flatten()(conv_4)
hidden = Dense(intermediate_dim, activation='relu')(flat)
z_mean = Dense(latent_dim)(hidden)
z_log_var = Dense(latent_dim)(hidden)
def sampling(args):
z_mean, z_log_var = args
epsilon = K.random_normal(shape=(K.shape(z_mean)[0], latent_dim),
mean=0., stddev=epsilon_std)
return z_mean + K.exp(z_log_var) * epsilon
# note that "output_shape" isn't necessary with the TensorFlow backend
# so you could write `Lambda(sampling)([z_mean, z_log_var])`
z = Lambda(sampling, output_shape=(latent_dim,))([z_mean, z_log_var])
# we instantiate these layers separately so as to reuse them later
decoder_hid = Dense(intermediate_dim, activation='relu')
decoder_upsample = Dense(filters * 14 * 14, activation='relu')
if K.image_data_format() == 'channels_first':
output_shape = (batch_size, filters, 14, 14)
else:
output_shape = (batch_size, 14, 14, filters)
decoder_reshape = Reshape(output_shape[1:])
decoder_deconv_1 = Conv2DTranspose(filters,
kernel_size=num_conv,
padding='same',
strides=1,
activation='relu')
decoder_deconv_2 = Conv2DTranspose(filters,
kernel_size=num_conv,
padding='same',
strides=1,
activation='relu')
if K.image_data_format() == 'channels_first':
output_shape = (batch_size, filters, 29, 29)
else:
output_shape = (batch_size, 29, 29, filters)
decoder_deconv_3_upsamp = Conv2DTranspose(filters,
kernel_size=(3, 3),
strides=(2, 2),
padding='valid',
activation='relu')
decoder_mean_squash = Conv2D(img_chns,
kernel_size=2,
padding='valid',
activation='sigmoid')
hid_decoded = decoder_hid(z)
up_decoded = decoder_upsample(hid_decoded)
reshape_decoded = decoder_reshape(up_decoded)
deconv_1_decoded = decoder_deconv_1(reshape_decoded)
deconv_2_decoded = decoder_deconv_2(deconv_1_decoded)
x_decoded_relu = decoder_deconv_3_upsamp(deconv_2_decoded)
x_decoded_mean_squash = decoder_mean_squash(x_decoded_relu)
# Custom loss layer
class CustomVariationalLayer(Layer):
def __init__(self, **kwargs):
self.is_placeholder = True
super(CustomVariationalLayer, self).__init__(**kwargs)
def vae_loss(self, x, x_decoded_mean_squash):
x = K.flatten(x)
x_decoded_mean_squash = K.flatten(x_decoded_mean_squash)
xent_loss = img_rows * img_cols * metrics.binary_crossentropy(x, x_decoded_mean_squash)
kl_loss = - 0.5 * K.mean(1 + z_log_var - K.square(z_mean) - K.exp(z_log_var), axis=-1)
return K.mean(xent_loss + kl_loss)
def call(self, inputs):
x = inputs[0]
x_decoded_mean_squash = inputs[1]
loss = self.vae_loss(x, x_decoded_mean_squash)
self.add_loss(loss, inputs=inputs)
# We don't use this output.
return x
y = CustomVariationalLayer()([x, x_decoded_mean_squash])
vae = Model(x, y)
vae.compile(optimizer='rmsprop', loss=None)
vae.summary()
# train the VAE on MNIST digits
(x_train, _), (x_test, y_test) = mnist.load_data()
x_train = x_train.astype('float32') / 255.
x_train = x_train.reshape((x_train.shape[0],) + original_img_size)
x_test = x_test.astype('float32') / 255.
x_test = x_test.reshape((x_test.shape[0],) + original_img_size)
print('x_train.shape:', x_train.shape)
vae.fit(x_train,
shuffle=True,
epochs=epochs,
batch_size=batch_size,
validation_data=(x_test, None))
# build a model to project inputs on the latent space
encoder = Model(x, z_mean)
# display a 2D plot of the digit classes in the latent space
x_test_encoded = encoder.predict(x_test, batch_size=batch_size)
plt.figure(figsize=(6, 6))
plt.scatter(x_test_encoded[:, 0], x_test_encoded[:, 1], c=y_test)
plt.colorbar()
plt.show()
# build a digit generator that can sample from the learned distribution
decoder_input = Input(shape=(latent_dim,))
_hid_decoded = decoder_hid(decoder_input)
_up_decoded = decoder_upsample(_hid_decoded)
_reshape_decoded = decoder_reshape(_up_decoded)
_deconv_1_decoded = decoder_deconv_1(_reshape_decoded)
_deconv_2_decoded = decoder_deconv_2(_deconv_1_decoded)
_x_decoded_relu = decoder_deconv_3_upsamp(_deconv_2_decoded)
_x_decoded_mean_squash = decoder_mean_squash(_x_decoded_relu)
generator = Model(decoder_input, _x_decoded_mean_squash)
# display a 2D manifold of the digits
n = 15 # figure with 15x15 digits
digit_size = 28
figure = np.zeros((digit_size * n, digit_size * n))
# linearly spaced coordinates on the unit square were transformed through the inverse CDF (ppf) of the Gaussian
# to produce values of the latent variables z, since the prior of the latent space is Gaussian
grid_x = norm.ppf(np.linspace(0.05, 0.95, n))
grid_y = norm.ppf(np.linspace(0.05, 0.95, n))
for i, yi in enumerate(grid_x):
for j, xi in enumerate(grid_y):
z_sample = np.array([[xi, yi]])
z_sample = np.tile(z_sample, batch_size).reshape(batch_size, 2)
x_decoded = generator.predict(z_sample, batch_size=batch_size)
digit = x_decoded[0].reshape(digit_size, digit_size)
figure[i * digit_size: (i + 1) * digit_size,
j * digit_size: (j + 1) * digit_size] = digit
plt.figure(figsize=(10, 10))
plt.imshow(figure, cmap='Greys_r')
plt.show()
参考文献:
[1] Keras Examples:https://github.com/fchollet/keras/tree/master/examples