-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathperceptual_distance.py
302 lines (267 loc) · 20.3 KB
/
perceptual_distance.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
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
import argparse
from pprint import pprint
import csv
from collections import defaultdict
import lpips
from torchvision import transforms
import PIL
from PIL import Image
import numpy as np
import medpy.io as medpy
def open_and_convert_image(impath, image_size):
if impath[-4:] == '.mha':
n = 5
img, _ = medpy.load(impath)
img = img[:,:,n]
img = Image.fromarray(img).convert("RGB")
else:
img = Image.open(impath).convert('RGB')
# rescale to image_size, convert PIL Image to torch tensor
transform = transforms.Compose([
transforms.Resize(image_size, interpolation=PIL.Image.BICUBIC),
transforms.CenterCrop(image_size),
transforms.ToTensor(),
])
im_torch = transform(img)
# rescale pixels from range [0, 1] (default for transforms.ToTensor()) to [-1, 1]
im_torch = 2*im_torch - 1
# output tensor should have size [1, C, H, W]
im_torch = im_torch.unsqueeze(0)
return im_torch
# We use the Perceptual Similarity Metric library,
# from the paper "The Unreasonable Effectiveness of Deep Features as a Perceptual Metric" (Zhang et al., 2018)
def perceptual_distance(im1, im2, device):
im1 = im1.to(device)
im2 = im2.to(device)
loss_fn_alex = lpips.LPIPS(net='alex').to(device)
loss_fn_vgg = lpips.LPIPS(net='vgg').to(device)
loss_fn_squeeze = lpips.LPIPS(net='squeeze').to(device)
d_alex = loss_fn_alex.forward(im1, im2)
d_vgg = loss_fn_vgg.forward(im1, im2)
d_squeeze = loss_fn_squeeze.forward(im1, im2)
return d_alex.cpu().detach().numpy().item(), d_vgg.cpu().detach().numpy().item(), d_squeeze.cpu().detach().numpy().item()
original_images = {
'bach' : ['iv001.tif', './sample_images/bach/iv001.tif'],
'chestx' : ['00000001_000.png', './sample_images/chestx/00000001_000.png'],
'chexpert' : ['patient00001_view1_frontal.jpg','./sample_images/chexpert/patient00001_view1_frontal.jpg'],
'diabetic_retinopathy' : ['34680_left.jpeg', './sample_images/diabetic_retinopathy/34680_left.jpeg'],
'ichallenge_amd' : ['AMD_A0001.jpg', './sample_images/ichallenge_amd/AMD_A0001.jpg'],
'ichallenge_pm' : ['H0009.jpg', './sample_images/ichallenge_pm/H0009.jpg'],
'montgomerycxr' : ['MCUCXR_0001_0.png', './sample_images/montgomerycxr/MCUCXR_0001_0.png'],
'shenzhencxr' : ['CHNCXR_0076_0.png', './sample_images/shenzhencxr/CHNCXR_0076_0.png'],
'stoic' : ['8622.mha', 'sample_images/stoic/8622.mha'],
'imagenet' : ['goldfish.jpeg', 'sample_images/imagenet/goldfish.jpeg']
}
reconstructed_images = {
'bach' : {'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_iv001.tif',
'swav': './reconstructed_images/swav/swav_True_iv001.tif',
'byol': './reconstructed_images/byol/byol_True_iv001.tif',
'pirl': './reconstructed_images/pirl/pirl_True_iv001.tif',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_iv001.tif',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_iv001.tif',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_iv001.tif',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_iv001.tif',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_iv001.tif',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_iv001.tif',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_iv001.tif',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_iv001.tif',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_iv001.tif'
},
'chestx' : {'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_00000001_000.png',
'swav': './reconstructed_images/swav/swav_True_00000001_000.png',
'byol': './reconstructed_images/byol/byol_True_00000001_000.png',
'pirl': './reconstructed_images/pirl/pirl_True_00000001_000.png',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_00000001_000.png',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_00000001_000.png',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_00000001_000.png',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_00000001_000.png',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_00000001_000.png',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_00000001_000.png',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_00000001_000.png',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_00000001_000.png',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_00000001_000.png'
},
'chexpert' : {'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_patient00001_view1_frontal.jpg',
'swav': './reconstructed_images/swav/swav_True_patient00001_view1_frontal.jpg',
'byol': './reconstructed_images/byol/byol_True_patient00001_view1_frontal.jpg',
'pirl': './reconstructed_images/pirl/pirl_True_patient00001_view1_frontal.jpg',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_patient00001_view1_frontal.jpg',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_patient00001_view1_frontal.jpg',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_patient00001_view1_frontal.jpg',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_patient00001_view1_frontal.jpg',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_patient00001_view1_frontal.jpg',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_patient00001_view1_frontal.jpg',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_patient00001_view1_frontal.jpg',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_patient00001_view1_frontal.jpg',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_patient00001_view1_frontal.jpg'
},
'diabetic_retinopathy' : {'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_34680_left.jpeg',
'swav': './reconstructed_images/swav/swav_True_34680_left.jpeg',
'byol': './reconstructed_images/byol/byol_True_34680_left.jpeg',
'pirl': './reconstructed_images/pirl/pirl_True_34680_left.jpeg',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_34680_left.jpeg',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_34680_left.jpeg',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_34680_left.jpeg',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_34680_left.jpeg',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_34680_left.jpeg',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_34680_left.jpeg',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_34680_left.jpeg',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_34680_left.jpeg',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_34680_left.jpeg'
},
'ichallenge_amd' : {'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_AMD_A0001.jpg',
'swav': './reconstructed_images/swav/swav_True_AMD_A0001.jpg',
'byol': './reconstructed_images/byol/byol_True_AMD_A0001.jpg',
'pirl': './reconstructed_images/pirl/pirl_True_AMD_A0001.jpg',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_AMD_A0001.jpg',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_AMD_A0001.jpg',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_AMD_A0001.jpg',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_AMD_A0001.jpg',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_AMD_A0001.jpg',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_AMD_A0001.jpg',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_AMD_A0001.jpg',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_AMD_A0001.jpg',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_AMD_A0001.jpg'
},
'ichallenge_pm' :{'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_H0009.jpg',
'swav': './reconstructed_images/swav/swav_True_H0009.jpg',
'byol': './reconstructed_images/byol/byol_True_H0009.jpg',
'pirl': './reconstructed_images/pirl/pirl_True_H0009.jpg',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_H0009.jpg',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_H0009.jpg',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_H0009.jpg',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_H0009.jpg',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_H0009.jpg',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_H0009.jpg',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_H0009.jpg',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_H0009.jpg',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_H0009.jpg'
},
'montgomerycxr' :{'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_MCUCXR_0001_0.png',
'swav': './reconstructed_images/swav/swav_True_MCUCXR_0001_0.png',
'byol': './reconstructed_images/byol/byol_True_MCUCXR_0001_0.png',
'pirl': './reconstructed_images/pirl/pirl_True_MCUCXR_0001_0.png',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_MCUCXR_0001_0.png',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_MCUCXR_0001_0.png',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_MCUCXR_0001_0.png',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_MCUCXR_0001_0.png',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_MCUCXR_0001_0.png',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_MCUCXR_0001_0.png',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_MCUCXR_0001_0.png',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_MCUCXR_0001_0.png',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_MCUCXR_0001_0.png'
},
'shenzhencxr' : {'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_CHNCXR_0076_0.png',
'swav': './reconstructed_images/swav/swav_True_CHNCXR_0076_0.png',
'byol': './reconstructed_images/byol/byol_True_CHNCXR_0076_0.png',
'pirl': './reconstructed_images/pirl/pirl_True_CHNCXR_0076_0.png',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_CHNCXR_0076_0.png',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_CHNCXR_0076_0.png',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_CHNCXR_0076_0.png',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_CHNCXR_0076_0.png',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_CHNCXR_0076_0.png',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_CHNCXR_0076_0.png',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_CHNCXR_0076_0.png',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_CHNCXR_0076_0.png',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_CHNCXR_0076_0.png'
},
'stoic' : {'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_8622.jpeg',
'swav': './reconstructed_images/swav/swav_True_8622.jpeg',
'byol': './reconstructed_images/byol/byol_True_8622.jpeg',
'pirl': './reconstructed_images/pirl/pirl_True_8622.jpeg',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_8622.jpeg',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_8622.jpeg',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_8622.jpeg',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_8622.jpeg',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_8622.jpeg',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_8622.jpeg',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_8622.jpeg',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_8622.jpeg',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_8622.jpeg'
},
'imagenet' : {'simclr-v1' : './reconstructed_images/simclr-v1/simclr-v1_True_goldfish.jpeg',
'swav': './reconstructed_images/swav/swav_True_goldfish.jpeg',
'byol': './reconstructed_images/byol/byol_True_goldfish.jpeg',
'pirl': './reconstructed_images/pirl/pirl_True_goldfish.jpeg',
'moco-v2': './reconstructed_images/moco-v2/moco-v2_True_goldfish.jpeg',
'mimic-chexpert_lr_0.01': './reconstructed_images/mimic-chexpert_lr_0.01/mimic-chexpert_lr_0.01_True_goldfish.jpeg',
'mimic-chexpert_lr_0.1': './reconstructed_images/mimic-chexpert_lr_0.1/mimic-chexpert_lr_0.1_True_goldfish.jpeg',
'mimic-chexpert_lr_1.0': './reconstructed_images/mimic-chexpert_lr_1.0/mimic-chexpert_lr_1.0_True_goldfish.jpeg',
'mimic-cxr_r18_lr_1e-4': './reconstructed_images/mimic-cxr_r18_lr_1e-4/mimic-cxr_r18_lr_1e-4_True_goldfish.jpeg',
'mimic-cxr_d121_lr_1e-4': './reconstructed_images/mimic-cxr_d121_lr_1e-4/mimic-cxr_d121_lr_1e-4_True_goldfish.jpeg',
'supervised_r50': './reconstructed_images/supervised_r50/supervised_r50_True_goldfish.jpeg',
'supervised_r18': './reconstructed_images/supervised_r18/supervised_r18_True_goldfish.jpeg',
'supervised_d121': './reconstructed_images/supervised_d121/supervised_d121_True_goldfish.jpeg'
},
}
if __name__ == "__main__":
parser = argparse.ArgumentParser(description='Compute and save perceptual distances between two images.')
# parser.add_argument('-im1', '--image-1', type=str, default='', help='path for image 1')
# parser.add_argument('-im2', '--image-2', type=str, default='', help='path for image 2')
parser.add_argument('-i', '--image-size', type=int, default=224, help='the size of the input images')
parser.add_argument('--device', type=str, default='cpu', help='CUDA or CPU (cuda | cpu)')
args = parser.parse_args()
pprint(args)
results_dict_alex = {}
results_dict_vgg = {}
results_dict_squeezenet = {}
csv_columns = ['model']
for dataset in original_images:
#im1_name = original_images[dataset][0]
im1_path = original_images[dataset][1]
im1 = open_and_convert_image(im1_path, args.image_size)
results_dict_alex[dataset] = {}
results_dict_vgg[dataset] = {}
results_dict_squeezenet[dataset] = {}
csv_columns.append(dataset)
for model in reconstructed_images[dataset]:
im2_path = reconstructed_images[dataset][model]
im2 = open_and_convert_image(im2_path, args.image_size)
d_alex, d_vgg, d_squeeze = perceptual_distance(im1, im2, args.device)
# results_dict[dataset][model] = {
# 'AlexNet' : d_alex,
# 'VGG' : d_vgg,
# 'SqueezeNet' : d_squeeze,
# }
# print(f'Perceptual distance between images {im1_name} and {im2_path}:')
# pprint(results_dict[dataset][model])
results_dict_alex[dataset][model] = d_alex
results_dict_vgg[dataset][model] = d_vgg
results_dict_squeezenet[dataset][model] = d_squeeze
# Flip the nested structure on the results for csv saving
flipped = defaultdict(dict)
for key, val in results_dict_alex.items():
for subkey, subval in val.items():
flipped[subkey][key] = subval
results_dict_alex = dict(flipped)
flipped = defaultdict(dict)
for key, val in results_dict_vgg.items():
for subkey, subval in val.items():
flipped[subkey][key] = subval
results_dict_vgg = dict(flipped)
flipped = defaultdict(dict)
for key, val in results_dict_squeezenet.items():
for subkey, subval in val.items():
flipped[subkey][key] = subval
results_dict_squeezenet = dict(flipped)
try:
with open('./results/perceptual-distance/alexnet_distances.csv', 'w') as csvfile:
writer = csv.DictWriter(csvfile, fieldnames=csv_columns)
writer.writeheader()
for model in results_dict_alex:
row = dict({'model': model}, **results_dict_alex[model])
writer.writerow(row)
with open('./results/perceptual-distance/vgg_distances.csv', 'w') as csvfile:
writer = csv.DictWriter(csvfile, fieldnames=csv_columns)
writer.writeheader()
for model in results_dict_vgg:
row = dict({'model': model}, **results_dict_vgg[model])
writer.writerow(row)
with open('./results/perceptual-distance/squeezenet_distances.csv', 'w') as csvfile:
writer = csv.DictWriter(csvfile, fieldnames=csv_columns)
writer.writeheader()
for model in results_dict_squeezenet:
row = dict({'model': model}, **results_dict_squeezenet[model])
writer.writerow(row)
except IOError:
print("I/O error")