-
Notifications
You must be signed in to change notification settings - Fork 0
/
trainer.py
276 lines (195 loc) · 8.52 KB
/
trainer.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
import os
import math
from decimal import Decimal
import utility
import random
import h5py
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.autograd import Variable
from tqdm import tqdm
import scipy.io as sio
from data import common
import numpy as np
from model.common import cov2pca, matrix_init
# from scipy.misc import imresize
# import model
class Trainer():
def __init__(self, args, loader, my_model, model_NLEst, model_KMEst, my_loss, ckp):
# freeze_support()
self.args = args
self.scale = args.scale
self.ckp = ckp
self.loader_train = loader.loader_train
self.loader_test = loader.loader_test
self.model = my_model
self.model_NLEst = model_NLEst
self.model_KMEst= model_KMEst
self.loss = my_loss
self.optimizer = utility.make_optimizer(args, self.model)
self.scheduler = utility.make_scheduler(args, self.optimizer)
self.loss_NLEst = my_loss
# args.lr = args.lr
self.optimizer_NLEst = utility.make_optimizer(args, self.model_NLEst)
self.scheduler_NLEst = utility.make_scheduler(args, self.optimizer_NLEst)
self.loss_KMEst = my_loss
self.optimizer_KMEst = utility.make_optimizer(args, self.model_KMEst)
self.scheduler_KMEst = utility.make_scheduler(args, self.optimizer_KMEst)
if self.args.load != '.':
self.optimizer.load_state_dict(
torch.load(os.path.join(ckp.dir, 'optimizer.pt'))
)
for _ in range(len(ckp.log)): self.scheduler.step()
self.error_last = 1e2
def train(self):
epoch = self.scheduler.last_epoch + 1
self.optimizer.step()
self.optimizer_NLEst.step()
self.optimizer_KMEst.step()
self.scheduler.step()
self.scheduler_NLEst.step()
self.scheduler_KMEst.step()
self.loss_NLEst.step()
self.loss_KMEst.step()
self.loss.step()
matrix = matrix_init()
V_pca_ = sio.loadmat('data/V.mat')
V_pca_ = V_pca_['V_pca']
V_pca = torch.from_numpy(V_pca_).float().cuda()
V_pca = V_pca.contiguous().view(15, 225, 1, 1)
lr = self.scheduler.get_last_lr()[0]
self.ckp.write_log(
'[Epoch {}]\tLearning rate: {:.2e}'.format(epoch, Decimal(lr))
)
self.loss.start_log()
self.model.train()
self.loader_train.dataset.num = self.loader_train.dataset.hr_data.shape[0]
print("Data pairs: {}".format(self.loader_train.dataset.num))
#
# self.model_NLEst.train()
# self.model_KMEst.train()
timer_data, timer_model = utility.timer(), utility.timer()
for batch, (lr, quality_factor, sigma, scale_factor, covmat, hr, _)\
in enumerate(self.loader_train):
lr, quality_factor, sigma, scale_factor, covmat, hr = \
self.prepare([lr, quality_factor, sigma, scale_factor, covmat, hr])
# print(scale_factor[0,0,0,0])
timer_data.hold()
timer_model.tic()
_, _, hei, wid = hr.data.size()
_, _, hei_l, wid_l = lr.data.size()
else:
print('Skip this batch {}! (Loss: {})'.format(
batch + 1, loss.item()
))
break
timer_model.hold()
if (batch + 1) % self.args.print_every == 0:
self.ckp.write_log('[{}/{}]\t{}\t{:.1f}+{:.1f}s'.format(
(batch + 1) * self.args.batch_size,
len(self.loader_train.dataset),
self.loss.display_loss(batch),
timer_model.release(),
timer_data.release()))
timer_data.tic()
self.loss.end_log(len(self.loader_train))
self.error_last = self.loss.log[-1, -1]
def test(self):
epoch = self.scheduler.last_epoch + 1
self.ckp.write_log('\nEvaluation:')
scale_list = self.args.scale
self.ckp.add_log(torch.zeros(1, len(scale_list)))
self.model.eval()
no_eval = 0
self.model_NLEst.eval()
self.model_KMEst.eval()
matrix = matrix_init()
V_pca_ = sio.loadmat('data/V.mat')
V_pca_ = V_pca_['V_pca']
V_pca = torch.from_numpy(V_pca_).float().cuda()
# V_pca = V_pca.t()
V_pca = V_pca.contiguous().view(15, 225, 1, 1)
timer_test = utility.timer()
with torch.no_grad():
best_psnr = 0
for idx_scale, scale in enumerate(scale_list):
eval_acc = 0
self.loader_test.dataset.set_scale(idx_scale)
tqdm_test = tqdm(self.loader_test, ncols=120)
for idx_img, (lr_, hr_, filename) in enumerate(tqdm_test):
filename = filename[0]
quality_factor = 90
no_eval = (hr_.nelement() == 1)
if no_eval:
[lr_] = self.prepare([lr_])
else:
lr_, hr_ = self.prepare([lr_, hr_])
_, _, hei, wid = lr_.data.size()
hei, wid = lr_.shape[2:]
quality_factor = (105.0 - quality_factor) / 255.0*torch.ones([1, 1, hei, wid]).float().cuda()
sf = scale / 16.0
scale_factor = torch.ones(1, 1, hei, wid).float().cuda() * sf
## Estimating noise level
sigma_est = self.model_NLEst(lr_, quality_factor, 0)
## Estimating kernel
ker_est = self.model_KMEst(lr_,
torch.cat((scale_factor, quality_factor, sigma_est), 1), 0)
ker_est = ker_est * ( scale ** 2)
ker_est = cov2pca(matrix.cuda(), V_pca, ker_est) ## convert cov matrix to PCA coff
hei, wid = hr_.shape[2:]
deg_map = torch.cat(
(quality_factor, sigma, scale_factor, ker_est), 1)
deg_map = F.interpolate(deg_map, [hei, wid], mode='bicubic')
lr_ = F.interpolate(lr_, [hei, wid], mode='bicubic')
sr = self.model(lr_, deg_map, idx_scale)
sr = utility.quantize(sr, self.args.rgb_range)
save_list = [sr]
if no_eval:
eval_acc += 0
else:
eval_acc += utility.calc_psnr(
sr, hr_, scale, self.args.rgb_range,
benchmark=self.loader_test.dataset.benchmark
)
if self.args.save_results:
self.ckp.save_results(filename, save_list, idx_img, scale)
self.ckp.log[-1, idx_scale] = eval_acc / len(self.loader_test)
if best_psnr < self.ckp.log[-1, idx_scale]:
is_best = True
best_psnr = self.ckp.log[-1, idx_scale]
else:
is_best = False
best = self.ckp.log.max(0)
self.ckp.write_log(
'[{} x{}]\tPSNR: {:.3f} (Best: {:.3f} @epoch {})'.format(
self.args.data_test,
scale,
self.ckp.log[-1, idx_scale],
best[0][idx_scale],
best[1][idx_scale] + 1
)
)
mean_st = self.ckp.log.mean(1)
best_mean = mean_st.max(0)
# print(best_mean)
# print(best_mean[1][0] + 1 == epoch)
self.ckp.write_log(
'Total time: {:.2f}s\n'.format(timer_test.toc()), refresh=True
)
if not self.args.test_only:
self.ckp.save(self, epoch, is_best)
# self.ckp.save(self, epoch, is_best=(best[1][0] + 1 == epoch))
def prepare(self, l, volatile=False):
device = torch.device('cpu' if self.args.cpu else 'cuda')
def _prepare(tensor):
if self.args.precision == 'half': tensor = tensor.half()
return tensor.to(device)
return [_prepare(_l) for _l in l]
def terminate(self):
if self.args.test_only:
self.test()
return True
else:
epoch = self.scheduler.last_epoch + 1
return epoch >= self.args.epochs