-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathhsi_denoising_gauss_iid.py
244 lines (227 loc) · 9.36 KB
/
hsi_denoising_gauss_iid.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
import os
# os.environ["CUDA_VISIBLE_DEVICES"] = "1"
import torch
import torch.nn as nn
import torch.optim as optim
import torchvision
import torchvision.transforms as transforms
import argparse
from utility import *
from hsi_setup import Engine, train_options, make_dataset
from utility import dataloaders_hsi_test ###modified###
if __name__ == '__main__':
"""Training settings"""
parser = argparse.ArgumentParser(
description='Hyperspectral Image Denoising (Gaussian Noise)')
opt = train_options(parser)
print(opt)
print(torch.cuda.device_count())
"""Setup Engine"""
engine = Engine(opt)
"""Dataset Setting"""
HSI2Tensor = partial(HSI2Tensor, use_2dconv=engine.get_net().use_2dconv)
common_transform_1 = lambda x: x
common_transform_2 = Compose([
partial(rand_crop, cropx=32, cropy=32),
])
target_transform = HSI2Tensor()
train_transform_0 = Compose([
AddNoise(50),
HSI2Tensor()
])
train_transform_1 = Compose([
AddNoise(10),
HSI2Tensor()
])
train_transform_2 = Compose([
AddNoise(30),
HSI2Tensor()
])
train_transform_3 = Compose([
AddNoise(50),
HSI2Tensor()
])
train_transform_4 = Compose([
AddNoise(70),
HSI2Tensor()
])
train_transform_5 = Compose([
AddNoise(90),
HSI2Tensor()
])
train_transform_6 = Compose([
AddNoiseList((10,30,50,70,90)),
HSI2Tensor()
])
'''
train_transform_1 = Compose([
AddNoise(50),
HSI2Tensor()
])
train_transform_2 = Compose([
AddNoiseBlind([10, 30, 50, 70]),
HSI2Tensor()
])
'''
print('==> Preparing data..')
icvl_64_31_TL_0 = make_dataset(
opt, train_transform_0,
target_transform, common_transform_1,opt.batchSize )
icvl_64_31_TL_1 = make_dataset(
opt, train_transform_1,
target_transform, common_transform_1, opt.batchSize)
icvl_64_31_TL_2 = make_dataset(
opt, train_transform_2,
target_transform, common_transform_1,opt.batchSize)
icvl_64_31_TL_3 = make_dataset(
opt, train_transform_3,
target_transform, common_transform_1, opt.batchSize)
# icvl_64_31_TL_4 = make_dataset(
# opt, train_transform_4,
# target_transform, common_transform_2, opt.batchSize*4)
icvl_64_31_TL_4 = make_dataset(
opt, train_transform_4,
target_transform, common_transform_1, opt.batchSize)
icvl_64_31_TL_5 = make_dataset(
opt, train_transform_5,
target_transform, common_transform_1, opt.batchSize)
icvl_64_31_TL_6 = make_dataset(
opt, train_transform_6,
target_transform, common_transform_1, opt.batchSize)
'''
icvl_64_31_TL_2 = make_dataset(
opt, train_transform_2,
target_transform, common_transform_2, 64)
'''
"""Test-Dev"""
###modified###
basefolder = opt.testroot
mat_names = ['icvl_dynamic_512_50','icvl_dynamic_512_70','icvl_dynamic_512_90']
#mat_names = ['icvl_512_30', 'icvl_512_50']
mat_loaders = []
for noise in (50,70,90):
test_path = os.path.join(basefolder, str(noise)+'/')
#print('noise: ',noise,end='')
mat_loaders.append(dataloaders_hsi_test.get_dataloaders([test_path],verbose=True,grey=False)['test'])
###modified###
#print(icvl_64_31_TL_0.__len__())
max=30*7
if icvl_64_31_TL_0.__len__()*opt.batchSize > 2000:
max_epoch = max//2
if_100 = 1
if_eval = 1
epoch_per_save = 1
testsize = 10
else:
max_epoch = max
if_100 = 0
if_eval = 1
epoch_per_save = 10
testsize = 5
print('max_epoch: ',max_epoch)
"""Main loop"""
base_lr = opt.lr
if_val_any = 1
if opt.resetepoch != -1:
engine.epoch = opt.resetepoch
while engine.epoch < max_epoch:
if if_100:
epoch = engine.epoch * 2
else:
epoch = engine.epoch
display_learning_rate(engine.optimizer)
np.random.seed() # reset seed per epoch, otherwise the noise will be added with a specific pattern
if epoch == 0:
adjust_learning_rate(engine.optimizer, opt.lr)
elif epoch == 10:
adjust_learning_rate(engine.optimizer, base_lr*0.1)
elif epoch == 20:
adjust_learning_rate(engine.optimizer, base_lr*0.01)
elif epoch % 30 == 0 and epoch >29 and epoch < max:
adjust_learning_rate(engine.optimizer, base_lr*0.1)
elif epoch % 30 == 14 and epoch >29 and epoch < max:
adjust_learning_rate(engine.optimizer, base_lr*0.01)
elif epoch == max:
adjust_learning_rate(engine.optimizer, base_lr*0.001)
'''
elif engine.epoch % 30 == 1 and engine.epoch != 1:
adjust_learning_rate(engine.optimizer, base_lr)
elif engine.epoch % 30 == 0 and engine.epoch != 0:
adjust_learning_rate(engine.optimizer, base_lr*0.01)
'''
#print(if_100)
if epoch < 30:
#engine.validate(mat_loaders[0], 'icvl-validate-50')
print("Training with unbindwise noise 50dB")
engine.train(icvl_64_31_TL_0)
if if_val_any:
engine.validate(mat_loaders[0], 'icvl-validate-50',testsize)###modified###
if if_eval:
engine.validate(mat_loaders[1], 'icvl-validate-70',testsize)###modified###
engine.validate(mat_loaders[2], 'icvl-validate-90',testsize)###modified###
#engine.validate(mat_loaders[1], 'icvl-validate-50')
elif epoch < 60:
print("Training with 10dB")
engine.train(icvl_64_31_TL_1)
if if_val_any:
engine.validate(mat_loaders[0], 'icvl-validate-50',testsize)###modified###
if if_eval:
engine.validate(mat_loaders[1], 'icvl-validate-70',testsize)###modified###
engine.validate(mat_loaders[2], 'icvl-validate-90',testsize)###modified###
#engine.validate(mat_loaders[0], 'icvl-validate-50')
#engine.validate(mat_loaders[0], 'icvl-validate-30')
#engine.validate(mat_loaders[1], 'icvl-validate-50')
elif epoch < 90:
print("Training with 30dB")
engine.train(icvl_64_31_TL_2)
if if_val_any:
engine.validate(mat_loaders[0], 'icvl-validate-50',testsize)###modified###
engine.validate(mat_loaders[1], 'icvl-validate-70',testsize)###modified###
if if_eval:
engine.validate(mat_loaders[2], 'icvl-validate-90',testsize)###modified###
elif epoch < 120:
print("Training with 50dB")
engine.train(icvl_64_31_TL_3)
if if_val_any:
engine.validate(mat_loaders[0], 'icvl-validate-50',testsize)###modified###
engine.validate(mat_loaders[1], 'icvl-validate-70',testsize)###modified###
engine.validate(mat_loaders[2], 'icvl-validate-90',testsize)###modified###
elif epoch < 150:
print("Training with 70dB")
engine.train(icvl_64_31_TL_4)
if if_val_any:
engine.validate(mat_loaders[0], 'icvl-validate-50',testsize)###modified###
engine.validate(mat_loaders[1], 'icvl-validate-70',testsize)###modified###
engine.validate(mat_loaders[2], 'icvl-validate-90',testsize)###modified###
elif epoch < 180:
print("Training with 90dB")
engine.train(icvl_64_31_TL_5)
if if_val_any:
engine.validate(mat_loaders[0], 'icvl-validate-50',testsize)###modified###
engine.validate(mat_loaders[1], 'icvl-validate-70',testsize)###modified###
engine.validate(mat_loaders[2], 'icvl-validate-90',testsize)###modified###
else:
print("Training with random noise")
engine.train(icvl_64_31_TL_6)
if engine.epoch == max_epoch and engine.epoch == 15*7:
testsize = 50
MSIQAs = []
MSIQAs.append(engine.validate_MSIQA(mat_loaders[0], 'icvl-validate-50',folder='nssnn_iid'))
MSIQAs.append(engine.validate_MSIQA(mat_loaders[0], 'icvl-validate-70',folder='nssnn_iid'))
MSIQAs.append(engine.validate_MSIQA(mat_loaders[0], 'icvl-validate-90',folder='nssnn_iid'))
print(" PSNR SSIM SAM")
print("50dB: %.4f %.4f %.4f"%( MSIQAs[0][0],MSIQAs[0][1],MSIQAs[0][2]))
print("70dB: %.4f %.4f %.4f"%( MSIQAs[1][0],MSIQAs[1][1],MSIQAs[1][2]))
print("90dB: %.4f %.4f %.4f"%( MSIQAs[2][0],MSIQAs[2][1],MSIQAs[2][2]))
else:
if if_val_any:
engine.validate(mat_loaders[0], 'icvl-validate-50',testsize)###modified###
engine.validate(mat_loaders[1], 'icvl-validate-70',testsize)###modified###
engine.validate(mat_loaders[2], 'icvl-validate-90',testsize)###modified###
print('\nLatest Result Saving...')
model_latest_path = os.path.join(engine.basedir, engine.prefix, 'model_latest.pth')
engine.save_checkpoint(
model_out_path=model_latest_path
)
if engine.epoch % epoch_per_save == 0:###modified###
engine.save_checkpoint()