forked from xjtu-mia/OCT-Multitask
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathterminal.py
197 lines (190 loc) · 8.59 KB
/
terminal.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
'''
oct_processing.py-->split_data-->main_multitask_for_tr_yifuyuan.py
'''
import argparse
import os
import numpy as np
import sys
import sys
sys.path.append(sys.path[0]+'/Core')
sys.path.append(sys.path[0]+'/Core/multitask_models')
sys.path.append(sys.path[0]+'/Core/BM3D_py')
parser = argparse.ArgumentParser()
'''
pre processing
'''
parser.add_argument('--preprocessing', action='store_true',
help="flatten and enhance of OCT image")
parser.add_argument('--dataset', default='yifuyuan', help="name of dataset, including yifuyuan, duke, retouch")
parser.add_argument('--oct_device', default='Cirrus', help='OCT设备名称（仅当dataset为retouch时有效）')
parser.add_argument('--root', default='./datasets', help='数据集的根目录')
parser.add_argument('--bm3d_img_exist', action='store_true', help='是否存在BM3D图像')
parser.add_argument('--wo_cvx', action='store_true', help='是否使用CVX')
parser.add_argument('--split_dataset', action='store_true',
help="split dataset for training and testing")
parser.add_argument('--merge_dataset', action='store_true',
help="merging yifuyuan and retouch training set for training")
parser.add_argument('--split_cross_valid', action='store_true',
help="split retouch dataset as k-fold cross valid form")
'''
training and testing
'''
parser.add_argument("--device_id", type=str, default='1',
help="which GPU to use")
parser.add_argument('--k', type=int, default=6,
help="k-fold validation")
parser.add_argument('--training', action='store_true',
help="training the model")
parser.add_argument('--testing', action='store_true',
help="testing the model")
parser.add_argument('--cross_valid', action='store_true',
help="training the model with cross validation")
parser.add_argument('--epoch', type=int, default=65, help='max epoch for training, 65 for yifuyuan and duke, 25 for retouch.')
parser.add_argument('--generate_pseudo', action='store_true',
help="testing on retouch dataset for generating pseudo labels")
parser.add_argument('--backbone', type=str, default='resnetv2',
help="selecting a backbone model, "
"including resnetv2, vgg, resnet, convnext, swintrans, shuffletrans, mpvit")
parser.add_argument("--log_name", type=str,
default=None,
help="giving a log name. If is None, the log name will generate by the program, eg:dataset_backbone_seed_123.")
parser.add_argument('--pretrain_path', type=str,
default='./datasets/yifuyuan/result/yifuyuan_resnetv2_seed_8830/weights_final.pth',
help='pre-trained weights path')
parser.add_argument('--seedlist', type=int, nargs='*', default=[8830], help='input: 8830 1024 64')
parser.add_argument('--k_size', type=int, default=51, help='kernel size of laplacian conv ')
args = parser.parse_args()
if args.preprocessing:
from Core import oct_preprocess
oct_preprocess.main(
dataset=args.dataset,
oct_device=args.oct_device,
root=args.root,
bm3d_img_exist=args.bm3d_img_exist,
wo_cvx=args.wo_cvx
)
if args.split_dataset:
from Core import split_data
split_data.main(
root=args.root,
dataset=args.dataset
)
if args.generate_pseudo:
from Core import generate_empty_label
from Core import generate_pseudo_label
if args.training or args.dataset != 'retouch':
raise ValueError('using testing operation for generating pseudo labels on retouch dataset!')
for seed in args.seedlist:
if not args.log_name:
log_name = 'test_' + args.dataset + '_' + args.backbone + '_seed_' + str(seed)
else:
log_name = args.log_name
generate_empty_label.main(root=args.root,
dataset=args.dataset,
oct_device=args.oct_device)
if args.testing:
from Core.main_multitask_for_te_retouch import run
run(device_id=args.device_id,
action='test',
dataset=args.dataset,
oct_device=args.oct_device,
seed=seed,
root=args.root,
log_name=log_name,
ckp_path=args.pretrain_path,
backbone=args.backbone)
generate_pseudo_label.main(
root=args.root,
dataset=args.dataset,
oct_device=args.oct_device,
log_name=log_name,
seed=seed)
if args.merge_dataset:
from Core import rename_yifuyuan2retouch
from Core import copy_dataset
save_dir = os.path.join(args.root, 'retouch', args.oct_device+'_yifuyuan', 'mix', 'all_train')
for i, f in enumerate(['train', 'val']):
rename_yifuyuan2retouch.main(in_root=os.path.join(args.root, 'yifuyuan', f),
out_root=os.path.join(save_dir, f),
label_folder='pseudo_label', index_num=25+i)
for n in ['flatted_IN_img_512', 'pseudo_label']:
copy_dataset.copytree(os.path.join(args.root, 'retouch', args.oct_device, 'preprocessing_data', n),
os.path.join(save_dir, 'train', n))
action = []
if args.training:
action += ['train']
if args.testing:
action += ['test']
for seed in args.seedlist:
if args.dataset == 'yifuyuan':
if action:
from Core.main_multitask_for_tr_yifuyuan import run
for a in action:
print('%sing the model' % a)
run(device_id=args.device_id,
action=a,
seed=seed,
root=args.root,
log_name=args.log_name,
backbone=args.backbone,
k_size=args.k_size)
elif args.dataset == 'duke':
if action:
from Core.main_multitask_for_tr_duke import run
for a in action:
print('%sing the model' % a)
run(device_id=args.device_id,
action=a,
seed=seed,
root=args.root,
log_name=args.log_name,
backbone=args.backbone)
elif args.dataset == 'retouch':
if args.cross_valid:
if action:
from Core.main_multitask_for_tr_retouch_cross_val import run
for i in range(args.k):
for a in action:
print('%sing the model' % a)
run(device_id=args.device_id,
id=i,
action=a,
epoch=args.epoch,
dataset=args.dataset,
oct_device=args.oct_device,
seed=seed,
root=args.root,
pretraining_path=args.pretrain_path,
log_name=args.log_name,
backbone=args.backbone)
else:
if action:
from Core.main_multitask_for_tr_retouch_yifuyuan import run
for a in action:
print('%sing the model' % a)
run(device_id=args.device_id,
action=a,
epoch=args.epoch,
dataset=args.dataset,
oct_device=args.oct_device,
seed=seed,
root=args.root,
pretraining_path=args.pretrain_path,
log_name=args.log_name,
backbone=args.backbone)
else:
raise ValueError('wrong dataset name! please select a dataset from: yifuyuan, duke and retouch.')
if args.split_cross_valid:
from Core import split_data_for_cross_val
from Core import rename_yifuyuan2retouch
indir = os.path.join(args.root, 'retouch', args.oct_device)
save_dir = os.path.join(args.root, 'retouch', args.oct_device+'_yifuyuan', 'mix/cross_valid')
for seed in args.seedlist:
for i in range(args.k):
split_data_for_cross_val.splitting(indir,
save_dir,
oct_device=args.oct_device,
seed=seed,
id = i)
rename_yifuyuan2retouch.main(out_root=os.path.join(save_dir, str(i), 'train'),
label_folder='pseudo_label',index_num=25)