forked from DengPingFan/Inf-Net
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtraining_&_testing_semiinfnet_+_multiclassunet.py
212 lines (155 loc) · 7.99 KB
/
training_&_testing_semiinfnet_+_multiclassunet.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
# -*- coding: utf-8 -*-
"""Training_&_Testing SemiInfNet + MulticlassUNet.ipynb
Automatically generated by Colaboratory.
Original file is located at
https://colab.research.google.com/drive/1bmII5H1AHL4MQDGr6vsOCDi-BFtD_j1m
**Pooyan Rezaeipour Lasaki**
e-mails:
**Mohsen Safaei**
e-mails: [email protected] & [email protected]
**Saeed Chamani**
e-mails: [email protected]
*Biomedical Engineering Department, School of Electrical Engineering, "Iran University of Science and Technology", Tehran*
*To find the files which you have uploaded in your "Google Drive"*
"""
from google.colab import drive
drive.mount('/content/gdrive')
"""*Unzipping the mentioned file which is named "Inf-Net-master"*"""
!unzip "/content/gdrive/MyDrive/Inf-Net-master.zip"
"""*For recognizing the module named "Code"*"""
cd '/content/Inf-Net-master'
import sys
sys.path.append('/content/Inf-Net-master')
"""*Training "MyTrain_MulClsLungInf_UNet" for Semi-Inf-Net + Multi-class UNet*
**Note** that if you run this file uninterruptedly after Semi-Inf-net(second method), you do not face a problem because the folder named "Snapshots" was created automatically.
But for running this file seperately(which we do below), you must create a folder in "Snapshots" which is named "save_weights",then create a folder in "save_weights" which is named "{}".
Also change the last line which is " save_path='Semi-Inf-Net_UNet' " to " save_path='{}' "
"""
# -*- coding: utf-8 -*-
"""Preview
Code for 'Inf-Net: Automatic COVID-19 Lung Infection Segmentation from CT Scans'
submit to Transactions on Medical Imaging, 2020.
First Version: Created on 2020-05-13 (@author: Ge-Peng Ji)
"""
import os
import numpy as np
import torch.optim as optim
from Code.utils.dataloader_MulClsLungInf_UNet import LungDataset
from torchvision import transforms
# from LungData import test_dataloader, train_dataloader # pls change batch_size
from torch.utils.data import DataLoader
from Code.model_lung_infection.InfNet_UNet import *
def train(epo_num, num_classes, input_channels, batch_size, lr, save_path):
train_dataset = LungDataset(
imgs_path='./Dataset/COVID-SemiSeg/Dataset/TrainingSet/MultiClassInfection-Train/Imgs/',
# NOTES: prior is borrowed from the object-level label of train split
pseudo_path='./Dataset/COVID-SemiSeg/Dataset/TrainingSet/MultiClassInfection-Train/Prior/',
label_path='./Dataset/COVID-SemiSeg/Dataset/TrainingSet/MultiClassInfection-Train/GT/',
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])]))
train_dataloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True, num_workers=4)
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
lung_model = Inf_Net_UNet(input_channels, num_classes) # input_channels=3， n_class=3
print(lung_model)
lung_model = lung_model.to(device)
criterion = nn.BCELoss().to(device)
optimizer = optim.SGD(lung_model.parameters(), lr=lr, momentum=0.7)
print("#" * 20, "\nStart Training (Inf-Net)\nThis code is written for 'Inf-Net: Automatic COVID-19 Lung "
"Infection Segmentation from CT Scans', 2020, TMI.\n"
"----\nPlease cite the paper if you use this code and dataset. "
"And any questions feel free to contact me "
"via E-mail ([email protected])\n----\n", "#" * 20)
for epo in range(epo_num):
train_loss = 0
lung_model.train()
for index, (img, pseudo, img_mask, _) in enumerate(train_dataloader):
img = img.to(device)
pseudo = pseudo.to(device)
img_mask = img_mask.to(device)
optimizer.zero_grad()
output = lung_model(torch.cat((img, pseudo), dim=1))
output = torch.sigmoid(output) # output.shape is torch.Size([4, 2, 160, 160])
loss = criterion(output, img_mask)
loss.backward()
iter_loss = loss.item()
train_loss += iter_loss
optimizer.step()
if np.mod(index, 20) == 0:
print('Epoch: {}/{}, Step: {}/{}, Train loss is {}'.format(epo, epo_num, index, len(train_dataloader), iter_loss))
os.makedirs('./checkpoints//UNet_Multi-Class-Semi', exist_ok=True)
if np.mod(epo+1, 10) == 0:
torch.save(lung_model.state_dict(),
'./Snapshots/save_weights/{}/unet_model_{}.pkl'.format(save_path, epo+1))
print('Saving checkpoints: unet_model_{}.pkl'.format(epo+1))
if __name__ == "__main__":
train(epo_num=200,
num_classes=3,
input_channels=6,
batch_size=16,
lr=1e-2,
save_path='{}')
"""*Testing "MyTest_MulClsLungInf_UNet" for Semi-Inf-Net + Multi-class UNet*
For running this file seperately from second method(which we do below), change the one before the last line which is " snapshot_dir='./Snapshots/save_weights/Semi-Inf-Net_UNet/unet_model_200.pkl' " to
" snapshot_dir='./Snapshots/save_weights/{}/unet_model_200.pkl' "
"""
# -*- coding: utf-8 -*-
"""Preview
Code for 'Inf-Net: Automatic COVID-19 Lung Infection Segmentation from CT Scans'
submit to Transactions on Medical Imaging, 2020.
First Version: Created on 2020-05-13 (@author: Ge-Peng Ji)
"""
import os
import numpy as np
from Code.utils.dataloader_MulClsLungInf_UNet import LungDataset
from torchvision import transforms
from torch.utils.data import DataLoader
from Code.model_lung_infection.InfNet_UNet import * # use U-Net for multi-class segmentation
from scipy import misc
from Code.utils.split_class import split_class
import shutil
def inference(num_classes, input_channels, snapshot_dir, save_path):
test_dataset = LungDataset(
imgs_path='./Dataset/COVID-SemiSeg/Dataset/TestingSet/MultiClassInfection-Test/Imgs/',
pseudo_path='./Dataset/COVID-SemiSeg/Results/Lung infection segmentation/Semi-Inf-Net/', # NOTES: generated from `Semi-Inf-Net`
label_path='./Dataset/COVID-SemiSeg/Dataset/TestingSet/MultiClassInfection-Test/GT/',
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])]),
is_test=True
)
test_dataloader = DataLoader(test_dataset, batch_size=1, shuffle=False, num_workers=0)
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
lung_model = Inf_Net_UNet(input_channels, num_classes).cuda()
print(lung_model)
lung_model.load_state_dict(torch.load(snapshot_dir))
lung_model.eval()
for index, (img, pseudo, img_mask, name) in enumerate(test_dataloader):
img = img.to(device)
pseudo = pseudo.to(device)
img_mask = img_mask.to(device)
output = lung_model(torch.cat((img, pseudo), dim=1))
output = torch.sigmoid(output) # output.shape is torch.Size([4, 2, 160, 160])
b, _, w, h = output.size()
_, _, w_gt, h_gt = img_mask.size()
# output b*n_class*h*w -- > b*h*w
pred = output.cpu().permute(0, 2, 3, 1).contiguous().view(-1, num_classes).max(1)[1].view(b, w, h).numpy().squeeze()
print('Class numbers of prediction in total:', np.unique(pred))
# pred = misc.imresize(pred, size=(w_gt, h_gt))
os.makedirs(save_path, exist_ok=True)
#misc.imsave(save_path + name[0].replace('.jpg', '.png'), pred)
import imageio
imageio.imwrite(save_path + name[0].replace('.jpg', '.png'), pred)
split_class(save_path, name[0].replace('.jpg', '.png'), w_gt, h_gt)
shutil.rmtree(save_path)
print('Test done!')
if __name__ == "__main__":
inference(num_classes=3,
input_channels=6,
snapshot_dir='./Snapshots/save_weights/{}/unet_model_200.pkl',
save_path='./Results/Multi-class lung infection segmentation/class_12/'
)