-
Notifications
You must be signed in to change notification settings - Fork 6
/
inpainting_metric.py
272 lines (224 loc) · 9.68 KB
/
inpainting_metric.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
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
import os
from glob import glob
import cv2
import lpips
import numpy as np
import torch
from scipy import linalg
from skimage.color import rgb2gray
try:
from skimage.measure import compare_ssim
except:
from skimage.metrics import structural_similarity as compare_ssim
from torch.autograd import Variable
from torch.nn.functional import adaptive_avg_pool2d
from tqdm import tqdm
from networks.inception import InceptionV3
def get_activations(images, model, batch_size=64, dims=2048,
cuda=False, verbose=False):
"""Calculates the activations of the pool_3 layer for all images.
Params:
-- images : Numpy array of dimension (n_images, 3, hi, wi). The values
must lie between 0 and 1.
-- model : Instance of inception model
-- batch_size : the images numpy array is split into batches with
batch size batch_size. A reasonable batch size depends
on the hardware.
-- dims : Dimensionality of features returned by Inception
-- cuda : If set to True, use GPU
-- verbose : If set to True and parameter out_step is given, the number
of calculated batches is reported.
Returns:
-- A numpy array of dimension (num images, dims) that contains the
activations of the given tensor when feeding inception with the
query tensor.
"""
model.eval()
d0 = images.shape[0]
if batch_size > d0:
print(('Warning: batch size is bigger than the data size. '
'Setting batch size to data size'))
batch_size = d0
n_batches = d0 // batch_size
if d0 % batch_size != 0:
n_batches += 1
n_used_imgs = d0
pred_arr = np.empty((n_used_imgs, dims))
with torch.no_grad():
for i in tqdm(range(n_batches)):
if verbose:
print('\rPropagating batch %d/%d' % (i + 1, n_batches),
end='', flush=True)
start = i * batch_size
end = min(start + batch_size, d0)
batch = torch.from_numpy(images[start:end]).type(torch.FloatTensor)
batch = Variable(batch)
if cuda:
batch = batch.cuda()
pred = model(batch)[0]
# If model output is not scalar, apply global spatial average pooling.
# This happens if you choose a dimensionality not equal 2048.
if pred.shape[2] != 1 or pred.shape[3] != 1:
pred = adaptive_avg_pool2d(pred, output_size=(1, 1))
pred_arr[start:end] = pred.cpu().data.numpy().reshape(end - start, -1)
if verbose:
print(' done')
return pred_arr
def calculate_frechet_distance(mu1, sigma1, mu2, sigma2, eps=1e-6):
"""Numpy implementation of the Frechet Distance.
The Frechet distance between two multivariate Gaussians X_1 ~ N(mu_1, C_1)
and X_2 ~ N(mu_2, C_2) is
d^2 = ||mu_1 - mu_2||^2 + Tr(C_1 + C_2 - 2*sqrt(C_1*C_2)).
Stable version by Dougal J. Sutherland.
Params:
-- mu1 : Numpy array containing the activations of a layer of the
inception net (like returned by the function 'get_predictions')
for generated samples.
-- mu2 : The sample mean over activations, precalculated on an
representive data set.
-- sigma1: The covariance matrix over activations for generated samples.
-- sigma2: The covariance matrix over activations, precalculated on an
representive data set.
Returns:
-- : The Frechet Distance.
"""
mu1 = np.atleast_1d(mu1)
mu2 = np.atleast_1d(mu2)
sigma1 = np.atleast_2d(sigma1)
sigma2 = np.atleast_2d(sigma2)
assert mu1.shape == mu2.shape, \
'Training and test mean vectors have different lengths'
assert sigma1.shape == sigma2.shape, \
'Training and test covariances have different dimensions'
diff = mu1 - mu2
# Product might be almost singular
covmean, _ = linalg.sqrtm(sigma1.dot(sigma2), disp=False)
if not np.isfinite(covmean).all():
msg = ('fid calculation produces singular product; '
'adding %s to diagonal of cov estimates') % eps
print(msg)
offset = np.eye(sigma1.shape[0]) * eps
covmean = linalg.sqrtm((sigma1 + offset).dot(sigma2 + offset))
# Numerical error might give slight imaginary component
if np.iscomplexobj(covmean):
if not np.allclose(np.diagonal(covmean).imag, 0, atol=1e-3):
m = np.max(np.abs(covmean.imag))
# raise ValueError('Imaginary component {}'.format(m))
covmean = covmean.real
tr_covmean = np.trace(covmean)
return (diff.dot(diff) + np.trace(sigma1) +
np.trace(sigma2) - 2 * tr_covmean)
def calculate_activation_statistics(images, model, batch_size=64,
dims=2048, cuda=False, verbose=False):
"""Calculation of the statistics used by the FID.
Params:
-- images : Numpy array of dimension (n_images, 3, hi, wi). The values
must lie between 0 and 1.
-- model : Instance of inception model
-- batch_size : The images numpy array is split into batches with
batch size batch_size. A reasonable batch size
depends on the hardware.
-- dims : Dimensionality of features returned by Inception
-- cuda : If set to True, use GPU
-- verbose : If set to True and parameter out_step is given, the
number of calculated batches is reported.
Returns:
-- mu : The mean over samples of the activations of the pool_3 layer of
the inception model.
-- sigma : The covariance matrix of the activations of the pool_3 layer of
the inception model.
"""
act = get_activations(images, model, batch_size, dims, cuda, verbose)
mu = np.mean(act, axis=0)
sigma = np.cov(act, rowvar=False)
return mu, sigma
def _compute_statistics_of_path(path, model, batch_size, dims, cuda):
npz_file = os.path.join(path, 'statistics.npz')
if os.path.exists(npz_file):
f = np.load(npz_file)
m, s = f['mu'][:], f['sigma'][:]
f.close()
else:
# path = pathlib.Path(path)
# files = list(path.glob('*.jpg')) + list(path.glob('*.png'))
files = list(glob(path + '/*.jpg')) + list(glob(path + '/*.png'))
files = sorted(files, key=lambda x: x.split('/')[-1])
imgs = []
for fn in files:
imgs.append(cv2.resize(cv2.imread(str(fn)), (299, 299), interpolation=cv2.INTER_LINEAR).astype(np.float32)[:, :, ::-1])
imgs = np.array(imgs)
# Bring images to shape (B, 3, H, W)
imgs = imgs.transpose((0, 3, 1, 2))
# Rescale images to be between 0 and 1
imgs /= 255
m, s = calculate_activation_statistics(imgs, model, batch_size, dims, cuda)
# np.savez(npz_file, mu=m, sigma=s)
return m, s
def calculate_fid_given_paths(paths, batch_size, cuda, dims):
"""Calculates the FID of two paths"""
# for p in paths:
# if not os.path.exists(p):
# raise RuntimeError('Invalid path: %s' % p)
block_idx = InceptionV3.BLOCK_INDEX_BY_DIM[dims]
model = InceptionV3([block_idx])
if cuda:
model.cuda()
print('calculate path1 statistics...')
m1, s1 = _compute_statistics_of_path(paths[0], model, batch_size, dims, cuda)
print('calculate path2 statistics...')
m2, s2 = _compute_statistics_of_path(paths[1], model, batch_size, dims, cuda)
print('calculate frechet distance...')
fid_value = calculate_frechet_distance(m1, s1, m2, s2)
return fid_value
def get_inpainting_metrics(src, tgt, get_fid=True):
input_paths = sorted(glob(src + '/*'), key=lambda x: x.split('/')[-1])
output_paths = sorted(glob(tgt + '/*'), key=lambda x: x.split('/')[-1])
assert len(input_paths) == len(output_paths), (len(input_paths), len(output_paths))
# PSNR and SSIM
psnrs = []
ssims = []
maes = []
mses = []
max_value = 1.0
for p1, p2 in tqdm(zip(input_paths, output_paths)):
img1 = cv2.imread(p1)
if img1 is None:
print(p1, 'is bad image!')
img2 = cv2.imread(p2)
if img2 is None:
print(p2, 'is bad image!')
if img1.shape[0] != img2.shape[0]:
img1 = cv2.resize(img1, (img2.shape[1], img2.shape[0]), interpolation=cv2.INTER_AREA)
try:
mse_ = np.mean((img1 / 255.0 - img2 / 255.0) ** 2)
except:
print(p1)
print(p2)
mae_ = np.mean(abs(img1 / 255.0 - img2 / 255.0))
psnr_ = max_value - 10 * np.log(mse_ + 1e-7) / np.log(10)
ssim_ = compare_ssim(rgb2gray(img1), rgb2gray(img2))
psnrs.append(psnr_)
ssims.append(ssim_)
mses.append(mse_)
maes.append(mae_)
psnr = np.mean(psnrs)
ssim = np.mean(ssims)
mse = np.mean(mses)
mae = np.mean(maes)
loss_fn_alex = lpips.LPIPS(net='alex').cuda()
with torch.no_grad():
ds = []
for im1, im2 in tqdm(zip(input_paths, output_paths)):
img1 = lpips.im2tensor(lpips.load_image(im1)).cuda()
img2 = lpips.im2tensor(lpips.load_image(im2)).cuda()
img2 = torch.nn.functional.interpolate(img2, size=(img1.shape[2], img1.shape[3]), mode='area')
d = loss_fn_alex(img1, img2)
ds.append(d)
ds = torch.stack(ds)
ds = torch.mean(ds).item()
# FID
if get_fid:
fid = calculate_fid_given_paths([src, tgt], batch_size=16, cuda=True, dims=2048)
return {'PSNR': psnr, 'SSIM': ssim, 'MSE': mse, 'MAE': mae, 'FID': fid, 'LPIPS': ds}
else:
return {'PSNR': psnr, 'SSIM': ssim, 'MSE': mse, 'MAE': mae, 'LPIPS': ds}