-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrain_fundus.py
281 lines (248 loc) · 12.3 KB
/
train_fundus.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
import os
import time
import datetime
import torch
import argparse
import numpy as np
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F
from tqdm import tqdm
from torchvision import transforms, models
from resnest.torch import resnest50
from net.SCNet.scnet import scnet50
from sklearn.metrics import cohen_kappa_score, f1_score, roc_auc_score, recall_score, precision_score, accuracy_score, \
hamming_loss
from data.base_dataset import Preproc, Rescale, RandomCrop, ToTensor, Normalization, Resize, ImgTrans
from data.csv_dataset import ImageDataset
# from torch.utils.tensorboard import SummaryWriter
from utils.utils import calc_kappa,hamming_score
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
cols = ['黄斑区视网膜出血', '黄斑区视网膜渗出', '黄斑区玻璃膜疣', '视网膜下橘红色病灶', '视网膜下出血']
classCount = len(cols)
data_dir = 'AMD_processed/'
list_dir = 'AMD_processed/label/new_two_stream/fundus/'
pre_models = \
{"resnet18": models.resnet18,
"resnet34": models.resnet34,
"resnet50": models.resnet50,
"resnest50": resnest50,
"scnet50": scnet50,
"inceptionv3": models.inception_v3,
"vgg16": models.vgg16,
"vgg19": models.vgg19,
"moco" : models.resnet50}
mean = {
224 : [0.485, 0.456, 0.406],
299 : [0.5, 0.5, 0.5]
}
std = {
224 : [0.229, 0.224, 0.225],
299 : [0.5, 0.5, 0.5]
}
def get_parser():
parser = argparse.ArgumentParser(description='Input hyperparameter of model:')
parser.add_argument('--root_path', type=str, default='/home/hejiawen/datasets',
help='The root path of dataset')
parser.add_argument('--fundus_model', type=str, default='resnet50',
choices=['resnet18', 'resnet34', 'resnet50', 'resnest50', 'scnet50', 'inceptionv3', 'vgg16',
'vgg19', 'moco'], help='The backbone model for Color fundus image')
parser.add_argument('--oct_model', type=str, default='resnet50',
choices=['resnet18', 'resnet34', 'resnet50', 'resnest50', 'scnet50', 'inceptionv3', 'vgg16',
'vgg19', 'moco'], help='The backbone model for OCT image')
parser.add_argument('--fundus_size', type=int, default=224, help='The input size for Color fundus image')
parser.add_argument('--oct_size', type=int, default=224, help='The input size for OCT image')
parser.add_argument('--epoch', type=int, default=500, help = 'The number of training epoch')
parser.add_argument('--batch_size', type=int, default=64, help='The size of batch')
parser.add_argument('--workers', type=int, default=1, help='The number of sub-processes to use for data loading')
parser.add_argument('--average', type=str, default='weighted',
choices=['micro', 'macro', 'weighted', 'samples'],
help='the type of averaging performed on the data')
parser.add_argument('--momentum', type=float, default=0.9, help='The momentum in optimizer')
parser.add_argument('--weight_decay', type=float, default=0.001, help='The weight_decay in optimizer')
parser.add_argument('--learning_rate', type=float, default=0.001, help='The learning_rate in optimizer')
parser.add_argument('--loss', type=str, default='bceloss', help='The loss function')
parser.add_argument('--use_gpu', type=int, default=0, choices=[0,1,2,3], help='The GPU on server used', required=True)
args = parser.parse_args()
return args
def pred2int(x):
out = []
for i in range(len(x)):
# print(x[i])
out.append([1 if y > 0.5 else 0 for y in x[i]])
return out
def train(model, train_loader, optimizer, scheduler, criterion, epoch, log):
print(f'Epoch={epoch}\n')
log.write(f'Epoch={epoch}\n')
model.train()
tbar = tqdm(train_loader, desc='\r', ncols=100)
y_pred = []
y_true = []
loss_val = 0
loss_valnorm = 0
# print(tbar)
for batch_idx, (inputs, target) in enumerate(tbar):
target = target.float()
data, target = inputs.cuda(), target.cuda()
optimizer.zero_grad()
output = model(data)
if 'inception' in args.fundus_model: #inception net 's output contains both hidden and output
new_output, hidden = output
output = new_output
loss = criterion(output, target)
loss.backward()
optimizer.step()
y_pred.extend(output.data.cpu().numpy())
y_true.extend(target.data.cpu().numpy())
tbar.set_description('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
epoch, batch_idx * len(inputs), len(train_loader.dataset),
100. * batch_idx / len(train_loader), loss.item()))
loss_val += loss.item()
loss_valnorm += 1
scheduler.step()
out_loss = loss_val / loss_valnorm
y_pred = np.array(y_pred)
y_true = np.array(y_true)
auroc = roc_auc_score(y_true, y_pred, average=args.average)
kappa = calc_kappa(y_true, y_pred, cols)
y_pred = pred2int(y_pred)
# print(y_pred)
# print(y_true)
f1 = f1_score(y_true, y_pred, average=args.average)
precision = precision_score(y_true, y_pred, average=args.average)
recall = recall_score(y_true, y_pred, average=args.average)
acc = accuracy_score(y_true=y_true, y_pred=y_pred)
hamming = hamming_loss(y_true, y_pred)
avg = (f1 + precision + recall + auroc) / 4.0
# writer.add_scalar("Train/f1", f1, epoch)
# writer.add_scalar("Train/auroc", auroc, epoch)f
# writer.add_scalar("Train/recall", recall, epoch)
# writer.add_scalar("Train/precision", precision, epoch)
# writer.add_scalar("Train/acc", acc, epoch)
# writer.add_scalar("Train/avg", avg, epoch)
# writer.add_scalar("Train/hamming_loss", hamming, epoch)
# writer.add_scalar("Train/ELoss", out_loss, epoch)
tbar.close()
print()
log.write('{:10s} {:10s} {:10s} {:10s} {:10s} {:10s} {:10s} {:10s}\n'.format('f1', 'auroc', 'recall', 'precision', 'acc', 'kappa', 'hamming', 'loss'))
log.write('{:10s} {:10s} {:10s} {:10s} {:10s} {:10s} {:10s} {:10s}\n'.format(str(round(f1,4)), str(round(auroc,4)), str(round(recall,4)), str(round(precision,4)), str(round(acc,4)), str(round(kappa,4)), str(round(hamming,4)), str(round(out_loss,4)) ))
# print(f1, auroc, recall, precision, acc, avg, hamming)
def validate(model, val_loader, criterion, epoch, log):
model.eval()
y_pred = []
y_true = []
tbar = tqdm(val_loader, desc='\r', ncols=100)
loss_val = 0
loss_valnorm = 0
with torch.no_grad():
for batch_idx, (inputs, target) in enumerate(tbar):
target = target.float()
data, target = inputs.cuda(), target.cuda()
output = model(data)
loss = criterion(output, target)
y_pred.extend(output.data.cpu().numpy())
y_true.extend(target.data.cpu().numpy())
# writer.add_scalar("Val/loss", loss.item(), epoch * len(val_loader) + batch_idx)
tbar.set_description('Val Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
epoch, batch_idx * len(inputs), len(val_loader.dataset),
100. * batch_idx / len(val_loader), loss.item()))
loss_val += loss.item()
loss_valnorm += 1
out_loss = loss_val / loss_valnorm
y_pred = np.array(y_pred)
y_true = np.array(y_true)
auroc = roc_auc_score(y_true, y_pred, average=args.average)
kappa = calc_kappa(y_true, y_pred, cols)
y_pred = pred2int(y_pred)
# sw = compute_sample_weight(class_weight='balanced', y=y_true)
f1 = f1_score(y_true, y_pred, average=args.average)
precision = precision_score(y_true, y_pred, average=args.average)
recall = recall_score(y_true, y_pred, average=args.average)
acc = accuracy_score(y_true=y_true, y_pred=y_pred)
hamming = hamming_loss(y_true=y_true, y_pred=y_pred)
avg = (f1 + recall + precision + auroc) / 4.0
# writer.add_scalar("Val/f1", f1, epoch)
# writer.add_scalar("Val/auroc", auroc, epoch)
# writer.add_scalar("Val/recall", recall, epoch)
# writer.add_scalar("Val/precision", precision, epoch)
# writer.add_scalar("Val/acc", acc, epoch)
# writer.add_scalar("Val/avg", avg, epoch)
# writer.add_scalar("Val/hamming_loss", hamming, epoch)
# writer.add_scalar("Val/ELoss", out_loss, epoch)
tbar.close()
print()
log.write('{:10s} {:10s} {:10s} {:10s} {:10s} {:10s} {:10s} {:10s}\n'.format('f1', 'auroc', 'recall', 'precision', 'acc', 'kappa', 'hamming', 'loss'))
log.write('{:10s} {:10s} {:10s} {:10s} {:10s} {:10s} {:10s} {:10s}\n'.format(str(round(f1,4)), str(round(auroc,4)), str(round(recall,4)), str(round(precision,4)), str(round(acc,4)), str(round(kappa,4)), str(round(hamming,4)), str(round(out_loss,4)) ))
# print(f1, auroc, recall, precision, acc, avg, hamming)
return avg
def main():
model = pre_models[args.fundus_model](pretrained=True)
if 'vgg' in args.fundus_model:
model.classifier[3] = nn.Linear(in_features=4096, out_features=1000, bias=True)
model.classifier[-1] = nn.Linear(in_features=1000, out_features=classCount, bias=True)
model.classifier.add_module(f'{len(model.classifier)}',nn.Sigmoid())
elif 'moco' in args.fundus_model:
checkpoint = torch.load('model/pre_train/moco_v2_800ep_pretrain.pth.tar')
weights = checkpoint['state_dict']
keys = list(weights.keys())
for key in keys:
if 'module.encoder_q' in key:
new_key = key.replace('module.encoder_q.', '')
weights[new_key] = weights[key]
del weights[key]
kernel_count = model.fc.weight.shape[1]
model.fc = nn.Sequential(nn.Linear(kernel_count, kernel_count), nn.ReLU(), nn.Linear(kernel_count, 128),\
nn.ReLU(), nn.Linear(128, classCount), nn.Sigmoid())
model.load_state_dict(weights, strict=False)
else:
kernel_count = model.fc.in_features
model.fc = nn.Sequential(nn.Linear(kernel_count, classCount), nn.Sigmoid())
train_tf = transforms.Compose([
Resize(args.fundus_size),
transforms.RandomHorizontalFlip(),
ToTensor(),
transforms.Normalize(mean=mean[args.fundus_size], std=std[args.fundus_size])
])
val_tf = transforms.Compose([
Resize(args.fundus_size),
ToTensor(),
transforms.Normalize(mean=mean[args.fundus_size], std=std[args.fundus_size])
])
train_loader = torch.utils.data.DataLoader(
ImageDataset(data_dir, 'train', train_tf, classCount, list_dir=list_dir),
batch_size=args.batch_size, shuffle=True,
num_workers=args.workers, pin_memory=True, drop_last=False
)
val_loader = torch.utils.data.DataLoader(
ImageDataset(data_dir, 'val', val_tf, classCount, list_dir=list_dir),
batch_size=args.batch_size, shuffle=False,
num_workers=args.workers, pin_memory=True, drop_last=False
)
model = model.cuda()
criterion = nn.BCELoss(reduction='mean')
optimizer = optim.SGD(model.parameters(), lr=args.learning_rate,
momentum=args.momentum,
weight_decay=args.weight_decay)
scheduler = optim.lr_scheduler.CosineAnnealingLR(optimizer, T_max=20, eta_min=0, last_epoch=-1)
train_log = open('logs/fundus/'+ model_name + '-train.log', 'w')
val_log = open('logs/fundus/'+ model_name + '-val.log', 'w')
max_avg = 0
for epoch in range(0, args.epoch):
train(model, train_loader, optimizer, scheduler, criterion, epoch, train_log)
avg = validate(model, val_loader, criterion, epoch, val_log)
if avg > max_avg:
torch.save(model, './model/fundus/' + model_name + '.pth')
max_avg = avg
if __name__ == '__main__':
args = get_parser()
NAME = str(args.epoch) + "+" + str(args.learning_rate) + '+' + str(args.weight_decay) + '+' + args.loss
model_name = datetime.datetime.now().strftime('%Y-%m-%d') + '+' + args.fundus_model + '+' + NAME
os.environ["CUDA_VISIBLE_DEVICES"] = str(args.use_gpu)
data_dir = os.path.join(args.root_path, data_dir)
list_dir = os.path.join(args.root_path, list_dir)
# writer = SummaryWriter(os.path.join('runs', 'fundus/' + model_name[:-4]))
print("Train fundus ", model_name)
start = time.time()
main()
end = time.time()
print('Finish fundus, Time=', end - start)