-
Notifications
You must be signed in to change notification settings - Fork 14
/
train_DA.py
254 lines (234 loc) · 13.7 KB
/
train_DA.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
import os
os.environ['OMP_NUM_THREADS'] = '1'
import sys
import shutil
from distutils.dir_util import copy_tree
import datetime
from tqdm import tqdm
import argparse
import numpy as np
import random
import torch
import torch.optim as optim
import torchvision.transforms as T
from torch.utils.data import DataLoader
from SFIT import datasets
from SFIT.models.classifier_shot import ClassifierShot, Discriminator
from SFIT.trainers import DATrainer
from SFIT.utils.str2bool import str2bool
from SFIT.utils.logger import Logger
def main(args):
# check if in debug mode
gettrace = getattr(sys, 'gettrace', None)
if gettrace():
print('Hmm, Big Debugger is watching me')
is_debug = True
else:
print('No sys.gettrace')
is_debug = False
# seed
if args.seed is not None:
np.random.seed(args.seed)
torch.manual_seed(args.seed)
torch.cuda.manual_seed(args.seed)
random.seed(args.seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
# torch.backends.cudnn.benchmark = True
else:
torch.backends.cudnn.benchmark = True
# dataset
data_path = os.path.expanduser(f'~/Data/{args.dataset}')
if args.dataset == 'digits':
n_classes = 10
use_src_test = True
args.batch_size = 64
if args.source == 'svhn' and args.target == 'mnist':
source_trans = T.Compose([T.Resize(32), T.ToTensor(), T.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])])
target_trans = T.Compose([T.Resize(32), T.Lambda(lambda x: x.convert("RGB")),
T.ToTensor(), T.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])])
source_train_dataset = datasets.SVHN(f'{data_path}/svhn', split='train', download=True,
transform=source_trans)
source_test_dataset = datasets.SVHN(f'{data_path}/svhn', split='test', download=True,
transform=source_trans)
target_train_dataset = datasets.MNIST(f'{data_path}/mnist', train=True, download=True,
transform=target_trans)
target_test_dataset = datasets.MNIST(f'{data_path}/mnist', train=False, download=True,
transform=target_trans)
args.arch = 'dtn'
elif args.source == 'usps' and args.target == 'mnist':
source_trans = T.Compose([T.RandomCrop(28, padding=4), T.RandomRotation(10),
T.ToTensor(), T.Normalize([0.5, ], [0.5, ])])
target_trans = T.Compose([T.ToTensor(), T.Normalize([0.5, ], [0.5, ])])
source_train_dataset = datasets.USPS(f'{data_path}/usps', train=True, download=True, transform=source_trans)
source_test_dataset = datasets.USPS(f'{data_path}/usps', train=False, download=True, transform=source_trans)
target_train_dataset = datasets.MNIST(f'{data_path}/mnist', train=True, download=True,
transform=target_trans)
target_test_dataset = datasets.MNIST(f'{data_path}/mnist', train=False, download=True,
transform=target_trans)
args.arch = 'lenet'
elif args.source == 'mnist' and args.target == 'usps':
source_trans = T.Compose([T.ToTensor(), T.Normalize([0.5, ], [0.5, ])])
target_trans = T.Compose([T.ToTensor(), T.Normalize([0.5, ], [0.5, ])])
source_train_dataset = datasets.MNIST(f'{data_path}/mnist', train=True, download=True,
transform=source_trans)
source_test_dataset = datasets.MNIST(f'{data_path}/mnist', train=False, download=True,
transform=source_trans)
target_train_dataset = datasets.USPS(f'{data_path}/usps', train=True, download=True, transform=target_trans)
target_test_dataset = datasets.USPS(f'{data_path}/usps', train=False, download=True, transform=target_trans)
args.arch = 'lenet'
else:
raise Exception('digits supports mnist, mnistm, usps, svhn')
elif args.dataset == 'office31':
n_classes = 31
use_src_test = False
args.epochs_S = 100
args.epochs_T = 15
if args.arch is None: args.arch = 'resnet50'
train_trans = T.Compose([T.Resize([256, 256]), T.RandomCrop(224), T.RandomHorizontalFlip(), T.ToTensor(),
T.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]), ])
test_trans = T.Compose([T.Resize([256, 256]), T.CenterCrop(224), T.ToTensor(),
T.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]), ])
source_train_dataset = datasets.ImageFolder(f'{data_path}/{args.source}/images', transform=train_trans)
source_test_dataset = datasets.ImageFolder(f'{data_path}/{args.source}/images', transform=train_trans)
target_train_dataset = datasets.ImageFolder(f'{data_path}/{args.target}/images', transform=train_trans)
target_test_dataset = datasets.ImageFolder(f'{data_path}/{args.target}/images', transform=test_trans)
elif args.dataset == 'visda':
n_classes = 12
use_src_test = False
args.lr_D *= 0.1
args.lr_S *= 0.1
args.lr_T *= 0.1
args.epochs_S = 10
args.epochs_T = 5
if args.arch is None: args.arch = 'resnet101'
args.source, args.target = 'syn', 'real'
train_trans = T.Compose([T.Resize([256, 256]), T.RandomCrop(224), T.RandomHorizontalFlip(), T.ToTensor(),
T.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]), ])
test_trans = T.Compose([T.Resize([256, 256]), T.CenterCrop(224), T.ToTensor(),
T.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5]), ])
source_train_dataset = datasets.ImageFolder(f'{data_path}/train', transform=train_trans)
source_test_dataset = datasets.ImageFolder(f'{data_path}/train', transform=train_trans)
target_train_dataset = datasets.ImageFolder(f'{data_path}/validation', transform=train_trans)
target_test_dataset = datasets.ImageFolder(f'{data_path}/validation', transform=test_trans)
else:
raise Exception('please choose dataset from [digits, office31, visda]')
if 'shot' in args.da_setting:
args.batch_size = 64
source_train_loader = DataLoader(source_train_dataset, batch_size=args.batch_size, shuffle=True,
num_workers=args.num_workers, drop_last=True)
source_train_loader_64 = DataLoader(source_train_dataset, batch_size=64, shuffle=True,
num_workers=args.num_workers, drop_last=True)
source_test_loader = DataLoader(source_test_dataset, batch_size=args.batch_size, shuffle=False,
num_workers=args.num_workers)
target_train_loader = DataLoader(target_train_dataset, batch_size=args.batch_size, shuffle=True,
num_workers=args.num_workers, drop_last=True)
target_test_loader = DataLoader(target_test_dataset, batch_size=args.batch_size, shuffle=False,
num_workers=args.num_workers)
logdir = f'logs/{args.da_setting}/{args.dataset}/s_{args.source}/t_{args.target}/' \
f'{"debug_" if is_debug else ""}{datetime.datetime.today():%Y-%m-%d_%H-%M-%S}/'
print(logdir)
# logging
if True:
os.makedirs(logdir + 'imgs', exist_ok=True)
copy_tree('./SFIT', logdir + 'scripts/SFIT')
for script in os.listdir('.'):
if script.split('.')[-1] == 'py':
dst_file = os.path.join(logdir, 'scripts', os.path.basename(script))
shutil.copyfile(script, dst_file)
sys.stdout = Logger(os.path.join(logdir, 'log.txt'), )
print('Settings:')
print(vars(args))
# model
net_D = Discriminator(args.bottleneck_dim).cuda()
net_S = ClassifierShot(n_classes, args.arch, args.bottleneck_dim, 'shot' in args.da_setting).cuda()
net_T = ClassifierShot(n_classes, args.arch, args.bottleneck_dim, 'shot' in args.da_setting).cuda()
# optimizers
optimizer_D = optim.SGD(net_D.parameters(), lr=args.lr_D, weight_decay=1e-3, momentum=0.9, nesterov=True)
if 'resnet' not in args.arch:
optimizer_S = optim.SGD(net_S.parameters(), lr=args.lr_S, weight_decay=1e-3, momentum=0.9, nesterov=True)
optimizer_T = optim.SGD(list(net_T.base.parameters()), # + list(net_T.bottleneck.parameters()),
lr=args.lr_T, weight_decay=1e-3, momentum=0.9, nesterov=True)
else:
optimizer_S = optim.SGD([{'params': net_S.base.parameters(), 'lr': args.lr_S * 0.1},
{'params': net_S.bottleneck.parameters()},
{'params': net_S.classifier.parameters()}],
lr=args.lr_S, weight_decay=1e-3, momentum=0.9, nesterov=True)
optimizer_T = optim.SGD([{'params': net_T.base.parameters(), 'lr': args.lr_T * 0.1}, ],
# {'params': net_T.bottleneck.parameters()}],
lr=args.lr_T, weight_decay=1e-3, momentum=0.9, nesterov=True)
# schedulers
scheduler_D = optim.lr_scheduler.CosineAnnealingWarmRestarts(optimizer_D, args.epochs_T, 1)
scheduler_S = optim.lr_scheduler.CosineAnnealingWarmRestarts(optimizer_S, args.epochs_S, 1)
scheduler_T = optim.lr_scheduler.CosineAnnealingWarmRestarts(optimizer_T, args.epochs_T, 1)
trainer = DATrainer(net_D, net_S, net_T, logdir, args.da_setting, args.source_LSR,
args.dataset == 'visda')
# source model
net_S_fpath = f'logs/{args.da_setting}/{args.dataset}/s_{args.source}/source_model.pth'
if os.path.exists(net_S_fpath) and not args.force_train_S:
print(f'Loading source model at: {net_S_fpath}...')
net_S.load_state_dict(torch.load(net_S_fpath))
pass
else:
print('Training source model...')
for epoch in tqdm(range(1, args.epochs_S + 1)):
trainer.train_net_S(epoch, source_train_loader_64, optimizer_S, scheduler_S)
if epoch % (max(args.epochs_S // 10, 1)) == 0:
if use_src_test:
print('Testing source model on [source]...')
trainer.test_net_S(source_test_loader)
print('Testing source model on [target]...')
trainer.test_net_S(target_test_loader)
torch.save(net_S.state_dict(), net_S_fpath)
torch.save(net_S.state_dict(), logdir + 'source_model.pth')
print('Testing source model on [source]...')
trainer.test_net_S(source_test_loader)
print('##############################################################')
print('Testing source model on [target]...')
print('##############################################################')
trainer.test_net_S(target_test_loader)
# target model & discriminator
net_T_fpath = f'logs/{args.da_setting}/{args.dataset}/s_{args.source}/t_{args.target}/target_model.pth'
print(f'Initialize target model with source model...')
net_T.load_state_dict(net_S.state_dict())
for epoch in tqdm(range(1, args.epochs_T + 1)):
print('Training target model...')
trainer.train_net_T(epoch, source_train_loader, target_train_loader, optimizer_T, optimizer_D,
[scheduler_T, scheduler_D])
if use_src_test:
print('Testing target model on [source]...')
trainer.test_net_T(source_test_loader)
print('Testing target model on [target]...')
trainer.test_net_T(target_test_loader)
torch.save(net_T.state_dict(), net_T_fpath)
torch.save(net_T.state_dict(), logdir + 'target_model.pth')
print('##############################################################')
print('Testing target model on [target]...')
print('##############################################################')
trainer.test_net_T(target_test_loader)
if __name__ == '__main__':
# settings
parser = argparse.ArgumentParser(description='Train SHOT')
parser.add_argument('-d', '--dataset', type=str, default='digits', choices=['digits', 'office31', 'visda'])
parser.add_argument('--source', type=str)
parser.add_argument('--target', type=str)
parser.add_argument('-a', '--arch', type=str, default=None,
choices=['alexnet', 'vgg16', 'resnet18', 'resnet50', 'digits'])
parser.add_argument('--bottleneck_dim', type=int, default=256)
parser.add_argument('-j', '--num_workers', type=int, default=4)
parser.add_argument('-b', '--batch-size', type=int, default=32, metavar='N',
help='input batch size for training (default: 64)')
parser.add_argument('--force_train_S', action='store_true', default=False)
# source model
parser.add_argument('--source_LSR', type=str2bool, default=True)
# target model
parser.add_argument('--da_setting', type=str, default='shot', choices=['shot', 'mmd', 'adda'])
parser.add_argument('--epochs_S', type=int, default=30, help='number of epochs to train')
parser.add_argument('--epochs_T', type=int, default=30, help='number of epochs to train')
parser.add_argument('--restart', type=float, default=1)
parser.add_argument('--lr_D', type=float, default=1e-3, help='discriminator learning rate')
parser.add_argument('--lr_S', type=float, default=1e-2, help='target model learning rate')
parser.add_argument('--lr_T', type=float, default=1e-2, help='source model learning rate')
parser.add_argument('--seed', type=int, default=None, help='random seed')
args = parser.parse_args()
main(args)