-
Notifications
You must be signed in to change notification settings - Fork 38
Expand file tree
/
Copy pathtrain_fully_supervised_2D.py
More file actions
207 lines (176 loc) · 8.66 KB
/
train_fully_supervised_2D.py
File metadata and controls
207 lines (176 loc) · 8.66 KB
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
import argparse
import logging
import os
import random
import shutil
import sys
import time
import numpy as np
import torch
import torch.backends.cudnn as cudnn
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from tensorboardX import SummaryWriter
from torch.nn import BCEWithLogitsLoss
from torch.nn.modules.loss import CrossEntropyLoss
from torch.utils.data import DataLoader
from torchvision import transforms
from torchvision.utils import make_grid
from tqdm import tqdm
from dataloaders import utils
from dataloaders.dataset import BaseDataSets, RandomGenerator
from networks.net_factory import net_factory
from utils import losses, metrics, ramps
from val_2D import test_single_volume, test_single_volume_ds
parser = argparse.ArgumentParser()
parser.add_argument('--root_path', type=str,
default='../data/ACDC', help='Name of Experiment')
parser.add_argument('--exp', type=str,
default='ACDC', help='experiment_name')
parser.add_argument('--fold', type=str,
default='fold5', help='cross validation')
parser.add_argument('--sup_type', type=str,
default='label', help='supervision type')
parser.add_argument('--model', type=str,
default='unet', help='model_name')
parser.add_argument('--num_classes', type=int, default=4,
help='output channel of network')
parser.add_argument('--max_iterations', type=int,
default=30000, help='maximum epoch number to train')
parser.add_argument('--batch_size', type=int, default=16,
help='batch_size per gpu')
parser.add_argument('--deterministic', type=int, default=1,
help='whether use deterministic training')
parser.add_argument('--base_lr', type=float, default=0.03,
help='segmentation network learning rate')
parser.add_argument('--patch_size', type=list, default=[256, 256],
help='patch size of network input')
parser.add_argument('--seed', type=int, default=2022, help='random seed')
args = parser.parse_args()
def train(args, snapshot_path):
base_lr = args.base_lr
num_classes = args.num_classes
batch_size = args.batch_size
max_iterations = args.max_iterations
model = net_factory(net_type=args.model, in_chns=1, class_num=num_classes)
db_train = BaseDataSets(base_dir=args.root_path, split="train", transform=transforms.Compose([
RandomGenerator(args.patch_size)
]), fold=args.fold, sup_type=args.sup_type)
db_val = BaseDataSets(base_dir=args.root_path,
fold=args.fold, split="val")
def worker_init_fn(worker_id):
random.seed(args.seed + worker_id)
trainloader = DataLoader(db_train, batch_size=batch_size, shuffle=True,
num_workers=8, pin_memory=True, worker_init_fn=worker_init_fn)
valloader = DataLoader(db_val, batch_size=1, shuffle=False,
num_workers=1)
model.train()
optimizer = optim.SGD(model.parameters(), lr=base_lr,
momentum=0.9, weight_decay=0.0001)
ce_loss = CrossEntropyLoss(ignore_index=4)
dice_loss = losses.DiceLoss(num_classes)
writer = SummaryWriter(snapshot_path + '/log')
logging.info("{} iterations per epoch".format(len(trainloader)))
iter_num = 0
max_epoch = max_iterations // len(trainloader) + 1
best_performance = 0.0
iterator = tqdm(range(max_epoch), ncols=70)
for epoch_num in iterator:
for i_batch, sampled_batch in enumerate(trainloader):
volume_batch, label_batch = sampled_batch['image'], sampled_batch['label']
volume_batch, label_batch = volume_batch.cuda(), label_batch.cuda()
outputs = model(volume_batch)
outputs_soft = torch.softmax(outputs, dim=1)
loss_ce = ce_loss(outputs, label_batch[:].long())
loss = 0.5 * (loss_ce + dice_loss(outputs_soft,
label_batch.unsqueeze(1)))
# loss = loss_ce
optimizer.zero_grad()
loss.backward()
optimizer.step()
lr_ = base_lr * (1.0 - iter_num / max_iterations) ** 0.9
for param_group in optimizer.param_groups:
param_group['lr'] = lr_
iter_num = iter_num + 1
writer.add_scalar('info/lr', lr_, iter_num)
writer.add_scalar('info/total_loss', loss, iter_num)
writer.add_scalar('info/loss_ce', loss_ce, iter_num)
logging.info(
'iteration %d : loss : %f, loss_ce: %f' %
(iter_num, loss.item(), loss_ce.item()))
if iter_num % 20 == 0:
image = volume_batch[1, 0:1, :, :]
image = (image - image.min()) / (image.max() - image.min())
writer.add_image('train/Image', image, iter_num)
outputs = torch.argmax(torch.softmax(
outputs, dim=1), dim=1, keepdim=True)
writer.add_image('train/Prediction',
outputs[1, ...] * 50, iter_num)
labs = label_batch[1, ...].unsqueeze(0) * 50
writer.add_image('train/GroundTruth', labs, iter_num)
if iter_num > 0 and iter_num % 200 == 0:
model.eval()
metric_list = 0.0
for i_batch, sampled_batch in enumerate(valloader):
metric_i = test_single_volume(
sampled_batch["image"], sampled_batch["label"], model, classes=num_classes)
metric_list += np.array(metric_i)
metric_list = metric_list / len(db_val)
for class_i in range(num_classes-1):
writer.add_scalar('info/val_{}_dice'.format(class_i+1),
metric_list[class_i, 0], iter_num)
writer.add_scalar('info/val_{}_hd95'.format(class_i+1),
metric_list[class_i, 1], iter_num)
performance = np.mean(metric_list, axis=0)[0]
mean_hd95 = np.mean(metric_list, axis=0)[1]
writer.add_scalar('info/val_mean_dice', performance, iter_num)
writer.add_scalar('info/val_mean_hd95', mean_hd95, iter_num)
if performance > best_performance:
best_performance = performance
save_mode_path = os.path.join(snapshot_path,
'iter_{}_dice_{}.pth'.format(
iter_num, round(best_performance, 4)))
save_best = os.path.join(snapshot_path,
'{}_best_model.pth'.format(args.model))
torch.save(model.state_dict(), save_mode_path)
torch.save(model.state_dict(), save_best)
logging.info(
'iteration %d : mean_dice : %f mean_hd95 : %f' % (iter_num, performance, mean_hd95))
model.train()
if iter_num % 3000 == 0:
save_mode_path = os.path.join(
snapshot_path, 'iter_' + str(iter_num) + '.pth')
torch.save(model.state_dict(), save_mode_path)
logging.info("save model to {}".format(save_mode_path))
if iter_num >= max_iterations:
break
if iter_num >= max_iterations:
iterator.close()
break
writer.close()
return "Training Finished!"
if __name__ == "__main__":
if not args.deterministic:
cudnn.benchmark = True
cudnn.deterministic = False
else:
cudnn.benchmark = False
cudnn.deterministic = True
random.seed(args.seed)
np.random.seed(args.seed)
torch.manual_seed(args.seed)
torch.cuda.manual_seed(args.seed)
snapshot_path = "../model/{}_{}/{}".format(
args.exp, args.fold, args.sup_type)
if not os.path.exists(snapshot_path):
os.makedirs(snapshot_path)
if os.path.exists(snapshot_path + '/code'):
shutil.rmtree(snapshot_path + '/code')
shutil.copytree('.', snapshot_path + '/code',
shutil.ignore_patterns(['.git', '__pycache__']))
logging.basicConfig(filename=snapshot_path+"/log.txt", level=logging.INFO,
format='[%(asctime)s.%(msecs)03d] %(message)s', datefmt='%H:%M:%S')
logging.getLogger().addHandler(logging.StreamHandler(sys.stdout))
logging.info(str(args))
train(args, snapshot_path)