-
Notifications
You must be signed in to change notification settings - Fork 21
/
utils.py
173 lines (139 loc) · 5.64 KB
/
utils.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
import os
from glob import glob
from tensorpack import *
from tensorpack.utils.viz import *
import tensorflow as tf
import numpy as np
SHAPE = 128
BATCH = 16
TEST_BATCH = 32
NF = 64 # channel size
def INReLU(x, name=None):
x = InstanceNorm('inorm', x)
return tf.nn.relu(x, name=name)
def INLReLU(x, name=None):
x = InstanceNorm('inorm', x)
return tf.nn.relu(x, name=name)
def BNLReLU(x, name):
x = BatchNorm('bn', x)
return tf.nn.relu(x)
def _tf_fspecial_gauss(size, sigma):
"""Function to mimic the 'fspecial' gaussian MATLAB function
"""
x_data, y_data = np.mgrid[-size//2 + 1:size//2 + 1, -size//2 + 1:size//2 + 1]
x_data = np.expand_dims(x_data, axis=-1)
x_data = np.expand_dims(x_data, axis=-1)
y_data = np.expand_dims(y_data, axis=-1)
y_data = np.expand_dims(y_data, axis=-1)
x = tf.constant(x_data, dtype=tf.float32)
y = tf.constant(y_data, dtype=tf.float32)
g = tf.exp(-((x**2 + y**2)/(2.0*sigma**2)))
return g / tf.reduce_sum(g)
def tf_ssim(img1, img2, cs_map=False, mean_metric=True, size=8, sigma=1.5):
window = _tf_fspecial_gauss(size, sigma) # window shape [size, size]
K1 = 0.03
K2 = 0.05
L = 1 # depth of image (255 in case the image has a differnt scale)
C1 = (K1*L)**2
C2 = (K2*L)**2
mu1 = tf.nn.conv2d(img1, window, strides=[1,1,1,1], padding='VALID')
mu2 = tf.nn.conv2d(img2, window, strides=[1,1,1,1], padding='VALID')
mu1_sq = mu1*mu1
mu2_sq = mu2*mu2
mu1_mu2 = mu1*mu2
sigma1_sq = tf.nn.conv2d(img1*img1, window, strides=[1,1,1,1],padding='VALID') - mu1_sq
sigma2_sq = tf.nn.conv2d(img2*img2, window, strides=[1,1,1,1],padding='VALID') - mu2_sq
sigma12 = tf.nn.conv2d(img1*img2, window, strides=[1,1,1,1],padding='VALID') - mu1_mu2
sigma1_sq = tf.abs(sigma1_sq)
sigma2_sq = tf.abs(sigma2_sq)
sigma12 = tf.abs(sigma12)
if cs_map:
value = (((2*mu1_mu2 + C1)*(2*sigma12 + C2))/((mu1_sq + mu2_sq + C1)*
(sigma1_sq + sigma2_sq + C2)),
(2.0*sigma12 + C2)/(sigma1_sq + sigma2_sq + C2))
else:
value = ((2*mu1_mu2 + C1)*(2*sigma12 + C2))/((mu1_sq + mu2_sq + C1)*
(sigma1_sq + sigma2_sq + C2))
if mean_metric:
value = tf.reduce_mean(value)
return value
def tf_ms_ssim(img1, img2, mean_metric=True, level=5):
#From NCHW to NHWC
img1 = tf.transpose(img1, [0, 2, 3, 1])
img2 = tf.transpose(img2, [0, 2, 3, 1])
weight = tf.constant([0.0448, 0.2856, 0.3001, 0.2363, 0.1333], dtype=tf.float32)
mssim = []
mcs = []
for l in range(level):
ssim_map, cs_map = tf_ssim(img1, img2, cs_map=True, mean_metric=False)
mssim.append(tf.reduce_mean(ssim_map))
mcs.append(tf.reduce_mean(cs_map))
filtered_im1 = tf.nn.avg_pool(img1, [1,2,2,1], [1,2,2,1], padding='SAME')
filtered_im2 = tf.nn.avg_pool(img2, [1,2,2,1], [1,2,2,1], padding='SAME')
img1 = filtered_im1
img2 = filtered_im2
# list to tensor of dim D+1
mssim = tf.stack(mssim, axis=0)
mcs = tf.stack(mcs, axis=0)
value = (tf.reduce_prod(mcs[0:level-1]**weight[0:level-1])*
(mssim[level-1]**weight[level-1]))
if mean_metric:
value = tf.reduce_mean(value)
return value
def tf_dssim(img1, img2):
img1 = tf.unstack(tf.expand_dims(img1, axis=2), axis=1)
img2 = tf.unstack(tf.expand_dims(img2, axis=2), axis=1)
value = tf.stack([tf_ms_ssim(i1, i2) for i1, i2 in zip(img1, img2)], axis=0)
return tf.subtract(1.0, tf.reduce_sum(value)/3, name='DSSIM_loss')
def get_celebA_data(datadir):
dfA = ImageFromFile(glob(datadir + "/trainA/*.jpg"), channel=3, shuffle=True)
dfB = ImageFromFile(glob(datadir + "/trainB/*.jpg"), channel=3, shuffle=True)
df = JoinData([dfA, dfB])
augs = [imgaug.Resize(SHAPE)]
df = AugmentImageComponents(df, augs, (0,1))
df = BatchData(df, BATCH)
df = PrefetchDataZMQ(df, 5)
return df
def get_data(datadir, isTrain=True):
if isTrain:
resize_range = (0.9, 1.1)
augs = [
imgaug.Flip(horiz=True),
imgaug.ResizeShortestEdge(int(SHAPE * 1.12)),
imgaug.Rotation(30),
imgaug.RandomCrop(int(SHAPE * 1.12)),
imgaug.RandomResize(resize_range, resize_range,
aspect_ratio_thres=0),
imgaug.RandomCrop(SHAPE),
]
else:
augs = [imgaug.ResizeShortestEdge(int(SHAPE * 1.12)),
imgaug.CenterCrop(SHAPE)
]
def get_image_pairs(dir1, dir2):
def get_df(dir):
files = sorted(glob(os.path.join(dir, '*.jpg')) +
glob(os.path.join(dir, '*.png')))
df = ImageFromFile(files, channel=3, shuffle=isTrain)
return AugmentImageComponent(df, augs)
return JoinData([get_df(dir1), get_df(dir2)])
names = ['trainA', 'trainB'] if isTrain else ['testA', 'testB']
df = get_image_pairs(*[os.path.join(datadir, n) for n in names])
df = BatchData(df, BATCH if isTrain else TEST_BATCH)
df = PrefetchDataZMQ(df, 8 if isTrain else 1)
return df
class VisualizeTestSet(Callback):
def __init__(self, data):
self.data = data
def _setup_graph(self):
self.pred = self.trainer.get_predictor(
['inputA', 'inputB'], ['viz_A_recon', 'viz_B_recon'])
def _before_train(self):
self.val_ds = get_data(self.data, isTrain=False)
def _trigger(self):
idx = 0
for iA, iB in self.val_ds.get_data():
vizA, vizB = self.pred(iA, iB)
self.trainer.monitors.put_image('testA-{}'.format(idx), vizA)
self.trainer.monitors.put_image('testB-{}'.format(idx), vizB)
idx += 1