-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrain.py
184 lines (163 loc) · 8.34 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
178
179
180
181
182
183
184
import argparse
import numpy as np
import logging
import torch
import torch.optim as optim
from torchvision import transforms
from retinanet import model
from retinanet.dataloader import CSVDataset, collater
from torch.utils.data import DataLoader
from retinanet.augmentor import VerticalFlip, HorizontalFlip, RandomNoise, Normalizer, ToTensor
from retinanet import csv_eval
def get_mean_std(loader):
channels_sum, channels_squared_sum, num_batches = 0, 0, 0
for data in loader:
img_data = data['img']
channels_sum += torch.mean(img_data, dim=[0,2,3])
channels_squared_sum += torch.mean(img_data**2, dim=[0,2,3])
num_batches += 1
mean = channels_sum / num_batches
std = (channels_squared_sum/num_batches - mean**2)**0.5
return mean, std
def main(args=None):
parser = argparse.ArgumentParser(description='Simple training script for training a RetinaNet network.')
parser.add_argument('--dataset', help='Dataset type, must be csv.')
parser.add_argument('--csv_train', help='Path to file containing training annotations (see readme)')
parser.add_argument('--csv_classes', help='Path to file containing class list (see readme)')
parser.add_argument('--csv_val', help='Path to file containing validation annotations (optional, see readme)')
parser.add_argument('--depth', help='Resnet depth, must be one of 18, 34, 50, 101, 152', type=int, default=50)
parser.add_argument('--epochs', help='Number of epochs', type=int, default=2)
parser = parser.parse_args(args)
# Create the data loaders
if parser.dataset == 'csv':
if parser.csv_train is None:
raise ValueError('Must provide --csv_train when training.')
if parser.csv_classes is None:
raise ValueError('Must provide --csv_classes when training.')
dataset_train = CSVDataset(train_file=parser.csv_train, class_list=parser.csv_classes,
transform=transforms.Compose([VerticalFlip(), HorizontalFlip(), RandomNoise(),
Normalizer(), ToTensor()]))
if parser.csv_val is None:
dataset_val = None
print('No validation annotations provided.')
else:
dataset_val = CSVDataset(train_file=parser.csv_val, class_list=parser.csv_classes,
transform=transforms.Compose([Normalizer(), ToTensor()]))
else:
raise ValueError('Dataset type not understood (must be csv or coco), exiting.')
# Instantiate dataloader
dataloader_train = DataLoader(dataset_train, batch_size=10, num_workers=3, shuffle=True, collate_fn=collater, drop_last=True)
if dataset_val is not None:
dataloader_val = DataLoader(dataset_val, batch_size=10, num_workers=3, shuffle=True, collate_fn=collater, drop_last=True)
"""
# Draw GT image and boxes
data = iter(dataloader_train).next()
img, annot = data["img"], data['annot']
img = img.numpy()
annot = annot.numpy()
for index, arg in enumerate(img):
for i in range(annot.shape[1]):
if -1 in annot[index][i]:
cv2.rectangle(img[index, 0], (1, 1), (1, 1), color=(0, 0, 255), thickness=2)
print(0,0,0,0)
else:
print(int(annot[index][i][0]), int(annot[index][i][1]), int(annot[index][i][2]), int(annot[index][i][3]))
cv2.rectangle(img[index, 0], (int(annot[index][i][0]), int(annot[index][i][1])), (int(annot[index][i][2]), int(annot[index][i][3])), color=(0, 0, 255), thickness=2)
plt.imshow(img[index, 0])
plt.show()
"""
# Create the model
if parser.depth == 18:
retinanet = model.resnet18(num_classes=dataset_train.num_classes(), pretrained=True)
elif parser.depth == 34:
retinanet = model.resnet34(num_classes=dataset_train.num_classes(), pretrained=True)
elif parser.depth == 50:
retinanet = model.resnet50(num_classes=dataset_train.num_classes(), pretrained=True)
elif parser.depth == 101:
retinanet = model.resnet101(num_classes=dataset_train.num_classes(), pretrained=True)
elif parser.depth == 152:
retinanet = model.resnet152(num_classes=dataset_train.num_classes(), pretrained=True)
else:
raise ValueError('Unsupported model depth, must be one of 18, 34, 50, 101, 152')
# Put model on GPU
use_gpu = True
if use_gpu:
if torch.cuda.is_available():
retinanet = retinanet.cuda()
retinanet = torch.nn.DataParallel(retinanet).cuda()
else:
retinanet = torch.nn.DataParallel(retinanet)
retinanet.training = True
optimizer = optim.Adam(retinanet.parameters(), lr=1e-5)
scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer, patience=3, verbose=True)
# loss_hist = collections.deque(maxlen=500)
print('Num training images: {}'.format(len(dataset_train)))
"""
# Determine mean and std across dataset
mean, std = get_mean_std(dataloader_train)
print("mean, std of cifar10:", mean, std)
"""
# Start training
global_mAP = 0.0
for epoch_num in range(parser.epochs):
retinanet.train()
retinanet.module.freeze_bn()
epoch_loss = []
avg_loss_per_epoch, avg_cls_loss_per_epoch, avg_reg_loss_per_epoch = 0.0, 0.0, 0.0
for iter_num, data in enumerate(dataloader_train):
try:
optimizer.zero_grad()
if torch.cuda.is_available():
classification_loss, regression_loss = retinanet([data['img'].cuda().float(), data['annot'].float()])
else:
classification_loss, regression_loss = retinanet([data['img'].float(), data['annot'].float()])
classification_loss = classification_loss.mean()
regression_loss = regression_loss.mean()
loss = classification_loss + regression_loss
# Accumulate losses
avg_loss_per_epoch += loss.item()
avg_cls_loss_per_epoch += classification_loss.item()
avg_reg_loss_per_epoch += regression_loss.item()
if bool(loss == 0):
continue
loss.backward()
# torch.nn.utils.clip_grad_norm_(retinanet.parameters(), 0.1)
optimizer.step()
epoch_loss.append(float(loss))
"""
print(
'Epoch: {} | Iteration: {} | Classification loss: {:1.5f} | Regression loss: {:1.5f} | Running loss: {:1.5f}'.format(
epoch_num, iter_num, float(classification_loss), float(regression_loss), np.mean(loss_hist)))
"""
del classification_loss
del regression_loss
except Exception as e:
print(e)
continue
scheduler.step(np.mean(epoch_loss))
# Calculate Average Loss and store the loss
avg_total_loss = avg_loss_per_epoch / len(dataloader_train)
avg_cls_loss = avg_cls_loss_per_epoch / len(dataloader_train)
avg_reg_loss = avg_reg_loss_per_epoch / len(dataloader_train)
print('Epoch: {} | Iteration: {} | Classification loss: {:1.5f} | Regression loss: {:1.5f} | Running loss: {:1.5f}'
.format(epoch_num, iter_num, float(avg_cls_loss), float(avg_reg_loss), avg_total_loss))
print("-" * 40)
# loss_hist.append(avg_total_loss)
logging.info('Average Training Loss (CE+RE): {}'.format(avg_total_loss))
if parser.dataset == 'csv' and parser.csv_val is not None:
print('Evaluating dataset')
logging.info('=================================')
mAP = csv_eval.evaluate(dataset_val, retinanet)
logging.info('mAP: {}'.format(mAP))
logging.info('=================================')
# Save weights if increase in mean average precision
if mAP[0][0] > global_mAP:
torch.save(retinanet.state_dict(), 'retinanet_lungNodule.pth')
global_mAP = mAP[0][0]
if __name__ == '__main__':
LOG_FILENAME = 'retinaNet.log'
logging.basicConfig(filename=LOG_FILENAME, level=logging.INFO, format='%(levelname)s: %(message)s')
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
logging.info(f'Using device {device}')
logging.info('------------------------------')
main()