-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmedsegmentation_ukagnet.py
202 lines (171 loc) · 7.1 KB
/
medsegmentation_ukagnet.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
import os
from copy import deepcopy
from glob import glob
import albumentations as A
import cv2
import hydra
import numpy as np
import torch.nn as nn
import torch.utils.data
from albumentations import Resize
from albumentations.augmentations import transforms
from albumentations.core.composition import Compose
from omegaconf import OmegaConf, open_dict
from sklearn.model_selection import train_test_split
from torchinfo import summary
from models import UKAGNet
from train import train_model, DiceLossWithBCE
class Dataset(torch.utils.data.Dataset):
def __init__(self, img_ids, img_dir, mask_dir, img_ext, mask_ext, num_classes, transform=None):
"""
Args:
img_ids (list): Image ids.
img_dir: Image file directory.
mask_dir: Mask file directory.
img_ext (str): Image file extension.
mask_ext (str): Mask file extension.
num_classes (int): Number of classes.
transform (Compose, optional): Compose transforms of albumentations. Defaults to None.
Note:
Make sure to put the files as the following structure:
<dataset name>
├── images
| ├── 0a7e06.jpg
│ ├── 0aab0a.jpg
│ ├── 0b1761.jpg
│ ├── ...
|
└── masks
├── 0
| ├── 0a7e06.png
| ├── 0aab0a.png
| ├── 0b1761.png
| ├── ...
|
├── 1
| ├── 0a7e06.png
| ├── 0aab0a.png
| ├── 0b1761.png
| ├── ...
...
"""
self.img_ids = img_ids
self.img_dir = img_dir
self.mask_dir = mask_dir
self.img_ext = img_ext
self.mask_ext = mask_ext
self.num_classes = num_classes
self.transform = transform
def __len__(self):
return len(self.img_ids)
def __getitem__(self, idx):
img_id = self.img_ids[idx]
img = cv2.imread(os.path.join(self.img_dir, img_id + self.img_ext))
mask = []
for i in range(self.num_classes):
# print(os.path.join(self.mask_dir, str(i),
# img_id + self.mask_ext))
mask.append(cv2.imread(os.path.join(self.mask_dir, str(i),
img_id + self.mask_ext), cv2.IMREAD_GRAYSCALE)[..., None])
mask = np.dstack(mask)
if self.transform is not None:
augmented = self.transform(image=img, mask=mask)
img = augmented['image']
mask = augmented['mask']
img = img.astype('float32') / 255
img = img.transpose(2, 0, 1)
mask = mask.astype('float32') / 255
mask = mask.transpose(2, 0, 1)
if mask.max() < 1:
mask[mask > 0] = 1.0
return img, mask
def get_data(dataset_name, data_dir, input_h, input_w, num_classes, seed):
img_ext = '.png'
if dataset_name == 'busi':
mask_ext = '_mask.png'
elif dataset_name == 'glas':
mask_ext = '.png'
elif dataset_name == 'cvc':
mask_ext = '.png'
# Data loading code
img_ids = sorted(glob(os.path.join(data_dir, dataset_name, 'images', '*' + img_ext)))
img_ids = [os.path.splitext(os.path.basename(p))[0] for p in img_ids]
train_img_ids, val_img_ids = train_test_split(img_ids, test_size=0.2, random_state=seed)
train_transform = Compose([
A.VerticalFlip(p=0.5),
A.HorizontalFlip(p=0.5),
A.RandomRotate90(p=0.5),
A.Transpose(p=0.5),
A.Rotate(p=0.5),
Resize(input_h, input_w),
transforms.Normalize(),
])
val_transform = Compose([
Resize(input_h, input_w),
transforms.Normalize(),
])
train_dataset = Dataset(
img_ids=train_img_ids,
img_dir=os.path.join(data_dir, dataset_name, 'images'),
mask_dir=os.path.join(data_dir, dataset_name, 'masks'),
img_ext=img_ext,
mask_ext=mask_ext,
num_classes=num_classes,
transform=train_transform)
val_dataset = Dataset(
img_ids=val_img_ids,
img_dir=os.path.join(data_dir, dataset_name, 'images'),
mask_dir=os.path.join(data_dir, dataset_name, 'masks'),
img_ext=img_ext,
mask_ext=mask_ext,
num_classes=num_classes,
transform=val_transform)
return train_dataset, val_dataset
@hydra.main(version_base=None, config_path="./configs/", config_name="medseg_ukans.yaml")
def main(cfg):
assert cfg.model.model_type in ['unet', 'u2net', 'u2net_small', 'vanilla_u2net',
'u2kagnet_bn'], "Unimplemented model"
model = None
for dataset_name, image_size in [("busi", 256), ("cvc", 256), ("glas", 512)]:
for mixer_type in ['conv', ]:
for use_bottleneck in [False, True]:
for ws in [4, 2, 1]:
if use_bottleneck:
model_type = f'bottleneck_ukagnet_{mixer_type}'
else:
model_type = f'ukagnet_{mixer_type}'
model = UKAGNet(
input_channels=3,
num_classes=1,
unet_depth=5,
unet_layers=2,
width_scale=1,
use_bottleneck=use_bottleneck,
mixer_type=mixer_type,
groups=cfg.model.groups,
degree=cfg.model.degree,
dropout=cfg.model.dropout,
affine=True,
norm_layer=nn.BatchNorm2d,
focal_window=3,
focal_level=2,
focal_factor=2,
use_postln_in_modulation=True,
normalize_modulator=True,
full_kan=True,
)
run_cfg = deepcopy(cfg)
OmegaConf.set_struct(run_cfg, True)
with open_dict(run_cfg):
run_cfg.wandb.runname = f"{model_type} x {ws}"
run_cfg.wandb.project_name = f'ukagnets-{dataset_name.lower()}'
run_cfg.output_dir = f"./experiments/{model_type}_{dataset_name.lower()}x{ws}/"
run_cfg.logging_dir = f"./experiments/{model_type}_{dataset_name.lower()}x{ws}/train_logs/"
run_cfg.model_name = f"{model_type}x{ws}"
summary(model, (4, 3, 256, 256), device='cuda:0')
dataset_train, dataset_val = get_data(dataset_name, "./data/UKAN_DATA/", image_size, image_size,
cfg.model.num_classes, cfg.seed)
loss_func = DiceLossWithBCE(smooth=1.0, bce_weight=0.1)
train_model(model, dataset_train, dataset_val, loss_func, run_cfg, dataset_test=None, cam_reporter=None)
if __name__ == '__main__':
main()