-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathtrain_gen.py
293 lines (232 loc) · 10.9 KB
/
train_gen.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
import time
import torch
import random
import numpy as np
import PIL
import torchvision
import os
import sys
import datetime
import pprint
import logging
import warnings
from options.train_options1 import TrainOptions
from torch.utils.data import DataLoader
from models import create_model
from cda.utils.utils import AverageMeter, create_logger
from pprint import pformat
from cda.config import cfg as detcfg
from cda.data.build import make_data_loader
from cda.utils.logger import setup_logger
from cda.utils import dist_util, mkdir, savefiles
from cda.utils.metric_logger import MetricLogger
warnings.filterwarnings("ignore")
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
basedir = os.path.dirname(os.path.realpath(__file__))
if __name__ == '__main__':
command_line = 'python ' + ' '.join(sys.argv)
opt = TrainOptions().parse()
seed = opt.seed
torch.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
torch.cuda.manual_seed(seed)
np.random.seed(seed)
random.seed(seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
detcfg.merge_from_file('./cda/config/{}_{}.yaml'.format(opt.train_classifier, opt.train_dataset))
opt.softmax2D = False
opt.data_dim = 'high'
opt.eps = 10.0 / 255.0 # CHANGED
opt.lr = 0.0002 # CHANGED
if not opt.pretrained_netG:
opt.pretrain_weights = ''
opt.train_classifier_weights = ''
detcfg.MODEL.BACKBONE.PRETRAINED = True
detcfg.MODEL.NUM_CLASSES = 1000
opt.classifier_weights = ''
opt.beta1 = 0.5
opt.lr_gamma = 0.3
eval_freq = opt.save_epoch_freq
opt.print_freq = 100
opt.save_latest_freq = 5000
opt.save_by_iter = True
# IMAGE SIZE
if detcfg.MODEL.META_ARCHITECTURE == 'inception_v3':
detcfg.INPUT.RESIZE_SIZE = 300
detcfg.INPUT.IMAGE_SIZE = 299
opt.train_getG_299 = True
else:
detcfg.INPUT.RESIZE_SIZE = 256
detcfg.INPUT.IMAGE_SIZE = 224
opt.train_getG_299 = False
detcfg.SOLVER.BATCH_SIZE = 16
detcfg.TEST.CONFIDENCE_THRESHOLD = 0.5
detcfg.OUTPUT_DIR = os.path.join(basedir, "checkpoints", "{}_{}_{}".format(
opt.train_dataset, opt.train_classifier, opt.loss_type))
detcfg.freeze()
if detcfg.OUTPUT_DIR:
mkdir(detcfg.OUTPUT_DIR)
opt.detcfg = detcfg
opt.detckpt = ''
opt.reqd_class_index = 0
opt.weightfile = ''
opt.continue_train = False
opt.load_iter = 6000
opt.attackobjective = 'Blind'
opt.input_nc = 3
opt.weight_L2 = 0
opt.weight_ce = 1
opt.weight_rl = 1
opt.weight_att = 1
opt.weight_feat = 1
opt.perturbmode = False
opt.stop_iter = 550000000
opt.pooling_type = 'Full'
opt.num_images = 5000
logger = setup_logger("CDA", dist_util.get_rank(), opt.detcfg.OUTPUT_DIR)
logger.info("Command line: {}".format(command_line))
logger.info("Experiment started at {}".format(datetime.datetime.now()
.strftime("%H:%M:%S secs on %d/%m/%y")))
logger.info("Environment:")
logger.info("\tPython: {}".format(sys.version.split(" ")[0]))
logger.info("\tPyTorch: {}".format(torch.__version__))
logger.info("\tTorchvision: {}".format(torchvision.__version__))
logger.info("\tCUDA: {}".format(torch.version.cuda))
logger.info("\tCUDNN: {}".format(torch.backends.cudnn.version()))
logger.info("\tNumPy: {}".format(np.__version__))
logger.info("\tPIL: {}".format(PIL.__version__))
logger.info(pformat(vars(opt)))
model = create_model(opt)
model.setup(opt)
dataloader = make_data_loader(opt.detcfg, is_train=True, distributed=False,
max_iter=1e+10, start_iter=0, shuffle=opt.data_shuffle,
data_aug=opt.data_aug)
opt.dataset = dataloader.dataset
total_iters = 0
avg_loss_ce = AverageMeter()
avg_loss_rl = AverageMeter()
avg_loss_att = AverageMeter()
avg_loss_L2 = AverageMeter()
avg_loss_feat = AverageMeter()
niter = 0
logger = logging.getLogger("CDA.trainer")
logger.info("Start training ...")
savefiles(detcfg.OUTPUT_DIR, opt)
num_iterations_per_epoch = int(len(dataloader.dataset) / detcfg.SOLVER.BATCH_SIZE)
logger.info("Number of iterations per epoch: {}, Total Images:{}".format(num_iterations_per_epoch,
len(dataloader.dataset)))
logger.info("epsilon: {:.1f}, seed:{}, warm start eps: {}, warm start L2 steps: {}".format(
opt.eps * 255, seed, opt.warm_start, opt.warm_start_L2_steps))
epoch_iter = 0
n_iter = 0
model.save_networks(0, detcfg.OUTPUT_DIR)
best_fooling = 0.0
best_model_epoch = 0
def eps_scheduler(epoch):
if epoch <= 3:
return 4.0 / 255.0
else:
return 10.0 / 255.0
flag = True
meters = MetricLogger()
end = time.time()
batch_time = time.time() - end
meters.update(time=batch_time)
for epoch in range(opt.epoch_count, opt.epoch_count + opt.max_epochs):
epoch_start_time = time.time()
iter_data_time = time.time()
end = time.time()
batch_time = time.time() - end
if epoch == 1 and opt.warm_start:
model.eps = eps_scheduler(epoch)
logger.info("Changing the epsilon to {:.1f}".format(model.eps * 255))
for i, data in enumerate(dataloader):
if 0 and total_iters % opt.print_freq == 0:
model.update_writer(total_iters, grad=False if i == 0 else True)
eta_seconds = meters.time.global_avg * (num_iterations_per_epoch - i)
eta_string = str(datetime.timedelta(seconds=int(eta_seconds)))
if total_iters > 2000 and flag:
model.eps = 10.0 / 255.0
logger.info("Changing the epsilon to {:.1f}".format(model.eps * 255))
flag = not flag
iter_start_time = time.time()
if total_iters % opt.print_freq == 0:
t_data = iter_start_time - iter_data_time
total_iters += 1
model.set_input(data)
batch_time = time.time() - end
end = time.time()
meters.update(time=batch_time)
if i == 0:
logger.info(model.image_ids[:2])
loss_dict = model.optimize_parameters()
avg_loss_ce.update(loss_dict['ce'].item())
avg_loss_rl.update(loss_dict['rl'].item())
avg_loss_att.update(loss_dict['att'].item())
avg_loss_L2.update(loss_dict['L2'].item())
avg_loss_feat.update(loss_dict['feat'].item())
if(niter % opt.print_freq == 0):
logger.info("epoch: {:3d}, iter: {:4d}, l2: {:.5f}, "
"ce: {:.5f}, rl: {:.5f}, att: {:.5f}, feat: {:.5f} eta: {}".format(epoch,
total_iters,
avg_loss_L2.avg,
avg_loss_ce.avg,
avg_loss_rl.avg,
avg_loss_att.avg,
avg_loss_feat.avg,
eta_string))
if total_iters % opt.display_freq == 0:
save_result = total_iters % opt.update_html_freq == 0
model.compute_visuals()
if total_iters % opt.print_freq == 0:
losses = model.get_current_losses()
t_comp = (time.time() - iter_start_time) / opt.batch_size
model.save_clean_and_adv(epoch)
if 1 and total_iters % opt.save_latest_freq == 0:
logger.info('saving the latest model (epoch {}, total_iters {})'.format(epoch,
total_iters))
save_suffix = 'iter_%d' % total_iters if opt.save_by_iter else 'latest'
model.save_networks(save_suffix, detcfg.OUTPUT_DIR)
model.evaluate_adv(total_iters, False)
iter_data_time = time.time()
if i == opt.stop_iter:
save_suffix = 'iter_%d' % total_iters if opt.save_by_iter else 'latest'
model.save_networks(save_suffix, detcfg.OUTPUT_DIR)
break
if i % 1500 == 0:
model.save_networks('latest', detcfg.OUTPUT_DIR)
if i == num_iterations_per_epoch:
n_iter = 0
break
niter += 1
if epoch % opt.save_epoch_freq == 0:
logger.info('saving the model at the end of epoch %d, iters %d' %
(epoch, total_iters))
model.save_networks('latest', detcfg.OUTPUT_DIR)
model.save_networks(epoch, detcfg.OUTPUT_DIR)
if epoch % eval_freq == 0:
adv_accu = model.evaluate_adv(total_iters, False)
if adv_accu['Fooling'] > best_fooling:
best_model_epoch = epoch
model.save_networks('best', detcfg.OUTPUT_DIR)
best_fooling = adv_accu['Fooling']
logger.info('saved the best model at the end of epoch: {} with fooling rate: {:.2f}%'.format(
best_model_epoch, 100 * best_fooling))
logger.info('saved the best model at the end of epoch: {} with fooling rate: {:.2f}%'.format(
best_model_epoch, 100 * best_fooling))
logger.info("End of epoch: {:3d}, iter: {:4d}, l2: {:.5f}, "
"ce: {:.5f}, rl: {:.5f}, att: {:.5f}, feat: {:.5f} eta: {}".format(epoch,
total_iters,
avg_loss_L2.avg,
avg_loss_ce.avg,
avg_loss_rl.avg,
avg_loss_att.avg,
avg_loss_feat.avg,
eta_string))
model.update_learning_rate()
torch.save({'state_dict': model.optimizer_G.state_dict()},
os.path.join(detcfg.OUTPUT_DIR, 'optimizer.pth'))
if epoch == opt.max_epochs:
logger.info("End of Training!")
break