-
Notifications
You must be signed in to change notification settings - Fork 26
/
Copy pathtrain.py
177 lines (142 loc) · 6.05 KB
/
train.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
# coding:utf-8
import os
import argparse
import time
import torch
import torch.nn.functional as F
from torch.autograd import Variable
from torch.utils.data import DataLoader
from util.MF_dataset import MF_dataset
from util.util import calculate_accuracy
from util.augmentation import RandomFlip, RandomCrop, RandomCropOut, RandomBrightness, RandomNoise
from model import MFNet, SegNet
from tqdm import tqdm
# config
n_class = 9
data_dir = '../../data/MF/'
model_dir = 'weights/'
augmentation_methods = [
RandomFlip(prob=0.5),
RandomCrop(crop_rate=0.1, prob=1.0),
# RandomCropOut(crop_rate=0.2, prob=1.0),
# RandomBrightness(bright_range=0.15, prob=0.9),
# RandomNoise(noise_range=5, prob=0.9),
]
lr_start = 0.01
lr_decay = 0.95
def train(epo, model, train_loader, optimizer):
lr_this_epo = lr_start * lr_decay**(epo-1)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_epo
loss_avg = 0.
acc_avg = 0.
start_t = t = time.time()
model.train()
for it, (images, labels, names) in enumerate(train_loader):
images = Variable(images).cuda(args.gpu)
labels = Variable(labels).cuda(args.gpu)
if args.gpu >= 0:
images = images.cuda(args.gpu)
labels = labels.cuda(args.gpu)
optimizer.zero_grad()
logits = model(images)
loss = F.cross_entropy(logits, labels)
loss.backward()
optimizer.step()
acc = calculate_accuracy(logits, labels)
loss_avg += float(loss)
acc_avg += float(acc)
cur_t = time.time()
if cur_t-t > 5:
print('|- epo %s/%s. train iter %s/%s. %.2f img/sec loss: %.4f, acc: %.4f' \
% (epo, args.epoch_max, it+1, train_loader.n_iter, (it+1)*args.batch_size/(cur_t-start_t), float(loss), float(acc)))
t += 5
content = '| epo:%s/%s lr:%.4f train_loss_avg:%.4f train_acc_avg:%.4f ' \
% (epo, args.epoch_max, lr_this_epo, loss_avg/train_loader.n_iter, acc_avg/train_loader.n_iter)
print(content)
with open(log_file, 'a') as appender:
appender.write(content)
def validation(epo, model, val_loader):
loss_avg = 0.
acc_avg = 0.
start_t = time.time()
model.eval()
with torch.no_grad():
for it, (images, labels, names) in enumerate(val_loader):
images = Variable(images)
labels = Variable(labels)
if args.gpu >= 0:
images = images.cuda(args.gpu)
labels = labels.cuda(args.gpu)
logits = model(images)
loss = F.cross_entropy(logits, labels)
acc = calculate_accuracy(logits, labels)
loss_avg += float(loss)
acc_avg += float(acc)
cur_t = time.time()
print('|- epo %s/%s. val iter %s/%s. %.2f img/sec loss: %.4f, acc: %.4f' \
% (epo, args.epoch_max, it+1, val_loader.n_iter, (it+1)*args.batch_size/(cur_t-start_t), float(loss), float(acc)))
content = '| val_loss_avg:%.4f val_acc_avg:%.4f\n' \
% (loss_avg/val_loader.n_iter, acc_avg/val_loader.n_iter)
print(content)
with open(log_file, 'a') as appender:
appender.write(content)
def main():
model = eval(args.model_name)(n_class=n_class)
if args.gpu >= 0: model.cuda(args.gpu)
optimizer = torch.optim.SGD(model.parameters(), lr=lr_start, momentum=0.9, weight_decay=0.0005)
# optimizer = torch.optim.Adam(model.parameters(), lr=lr_start)
if args.epoch_from > 1:
print('| loading checkpoint file %s... ' % checkpoint_model_file, end='')
model.load_state_dict(torch.load(checkpoint_model_file, map_location={'cuda:0':'cuda:1'}))
optimizer.load_state_dict(torch.load(checkpoint_optim_file))
print('done!')
train_dataset = MF_dataset(data_dir, 'train', have_label=True, transform=augmentation_methods)
val_dataset = MF_dataset(data_dir, 'val', have_label=True)
train_loader = DataLoader(
dataset = train_dataset,
batch_size = args.batch_size,
shuffle = True,
num_workers = args.num_workers,
pin_memory = True,
drop_last = True
)
val_loader = DataLoader(
dataset = val_dataset,
batch_size = args.batch_size,
shuffle = False,
num_workers = args.num_workers,
pin_memory = True,
drop_last = False
)
train_loader.n_iter = len(train_loader)
val_loader.n_iter = len(val_loader)
for epo in tqdm(range(args.epoch_from, args.epoch_max+1)):
print('\n| epo #%s begin...' % epo)
train(epo, model, train_loader, optimizer)
validation(epo, model, val_loader)
# save check point model
print('| saving check point model file... ', end='')
torch.save(model.state_dict(), checkpoint_model_file)
torch.save(optimizer.state_dict(), checkpoint_optim_file)
print('done!')
os.rename(checkpoint_model_file, final_model_file)
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Train MFNet with pytorch')
parser.add_argument('--model_name', '-M', type=str, default='MFNet')
parser.add_argument('--batch_size', '-B', type=int, default=8)
parser.add_argument('--epoch_max' , '-E', type=int, default=100)
parser.add_argument('--epoch_from', '-EF', type=int, default=1)
parser.add_argument('--gpu', '-G', type=int, default=0)
parser.add_argument('--num_workers', '-j', type=int, default=8)
args = parser.parse_args()
model_dir = os.path.join(model_dir, args.model_name)
os.makedirs(model_dir, exist_ok=True)
checkpoint_model_file = os.path.join(model_dir, 'tmp.pth')
checkpoint_optim_file = os.path.join(model_dir, 'tmp.optim')
final_model_file = os.path.join(model_dir, 'final.pth')
log_file = os.path.join(model_dir, 'log.txt')
print('| training %s on GPU #%d with pytorch' % (args.model_name, args.gpu))
print('| from epoch %d / %s' % (args.epoch_from, args.epoch_max))
print('| model will be saved in: %s' % model_dir)
main()