-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdataset.py
181 lines (138 loc) · 6.51 KB
/
dataset.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
import torch
import random
import numpy as np
from pre_processing import *
import torch.nn as nn
from random import randint
from PIL import Image, ImageSequence
import glob
from torch.utils.data.dataset import Dataset
BATCH_SIZE = 2
IN_SIZE = 1024
OUT_SIZE = 1024
TRAIN_VALID_RATIO = 0.8
# train_index = random.sample(range(0, 30), 24)
# test_index = list(set([i for i in range(0, 30)]) - set(train_index))
# print(train_index)
# print(test_index)
def CREMIDataTrain(image_path, mask_path, in_size=IN_SIZE, out_size=OUT_SIZE):
image_arr = Image.open(str(image_path))
mask_arr = Image.open(str(mask_path))
img_as_np = []
orig_img_as_np = []
for i, img_as_img in enumerate(ImageSequence.Iterator(image_arr)):
if i not in [idx for idx in range(80, 100)]:
singleImage_as_np = np.asarray(img_as_img)
img_as_np.append(singleImage_as_np)
orig_img_as_np.append(singleImage_as_np)
msk_as_np = []
orig_msk_as_np = []
for j, label_as_img in enumerate(ImageSequence.Iterator(mask_arr)):
if j not in [idx for idx in range(80, 100)]:
singleLabel_as_np = np.asarray(label_as_img)
msk_as_np.append(singleLabel_as_np)
orig_msk_as_np.append(singleLabel_as_np)
img_as_np, orig_img_as_np = np.stack(img_as_np, axis=0), np.stack(orig_img_as_np, axis=0)
msk_as_np, orig_msk_as_np = np.stack(msk_as_np, axis=0), np.stack(orig_msk_as_np, axis=0)
img_as_np, msk_as_np = flip(img_as_np, msk_as_np)
# Noise Determine {0: Gaussian_noise, 1: uniform_noise
if randint(0, 1):
gaus_sd, gaus_mean = randint(0, 20), 0
img_as_np = add_gaussian_noise(img_as_np, gaus_mean, gaus_sd)
else:
l_bound, u_bound = randint(-20, 0), randint(0, 20)
img_as_np = add_uniform_noise(img_as_np, l_bound, u_bound)
# change brightness
pix_add = randint(-20, 20)
img_as_np = change_brightness(img_as_np, pix_add)
img_as_np, orig_img_as_np = normalization2(img_as_np.astype(float), max=1, min=0), normalization2(
orig_img_as_np.astype(float), max=1, min=0)
# print(msk_as_np[0])
msk_as_np, orig_msk_as_np = msk_as_np / 255, orig_msk_as_np / 255
img_as_tensor = torch.from_numpy(img_as_np).float()
msk_as_tensor = torch.from_numpy(msk_as_np).long()
orig_img_as_tensor, orig_msk_as_tensor = torch.from_numpy(orig_img_as_np).float(), torch.from_numpy(
orig_msk_as_np).long()
img_as_tensor = torch.cat((img_as_tensor, orig_img_as_tensor), 0)
msk_as_tensor = torch.cat((msk_as_tensor, orig_msk_as_tensor), 0)
return (img_as_tensor, msk_as_tensor)
class ComDataset(Dataset):
def __init__(self, dataToSlice, SLICES_COLLECT):
self.n_slices = len(SLICES_COLLECT)
self.data = dataToSlice
# for i, data in enumerate(dataToSlice):
# self.data.append(data)
def __getitem__(self, index):
if self.n_slices == 1:
return self.data[0][index]
elif self.n_slices == 2:
return self.data[0][index], self.data[1][index]
elif self.n_slices == 3:
return self.data[0][index], self.data[1][index], self.data[2][index]
def __len__(self):
return len(self.data[0])
def CREMIDataVal(image_path, mask_path, in_size=IN_SIZE, out_size=OUT_SIZE):
image_arr = Image.open(str(image_path))
mask_arr = Image.open(str(mask_path))
img_as_np = []
for i, img_as_img in enumerate(ImageSequence.Iterator(image_arr)):
if i in [idx for idx in range(80, 100)]:
singleImage_as_np = np.asarray(img_as_img)
img_as_np.append(singleImage_as_np)
msk_as_np = []
for j, label_as_img in enumerate(ImageSequence.Iterator(mask_arr)):
if j in [idx for idx in range(80, 100)]:
singleLabel_as_np = np.asarray(label_as_img)
msk_as_np.append(singleLabel_as_np)
img_as_np = np.stack(img_as_np, axis=0)
msk_as_np = np.stack(msk_as_np, axis=0)
# Normalize
img_as_np = normalization2(img_as_np.astype(float), max=1, min=0)
msk_as_np = msk_as_np / 255
img_as_tensor = torch.from_numpy(img_as_np).float()
msk_as_tensor = torch.from_numpy(msk_as_np).long()
return (img_as_tensor, msk_as_tensor)
class CREMIDataPreTrained(Dataset):
def __init__(self, image_path, in_size=IN_SIZE, out_size=OUT_SIZE):
self.lh_images_array = glob.glob(str(image_path) + "/epoch_30/*lh.png")
self.lh_images_array = sorted([x.replace('lh.png', '') for x in self.lh_images_array])
self.binary_images_array = sorted(glob.glob(str(image_path) + "/epoch_30/*[0-9].png"))
self.binary_images_array = sorted([x.replace('.png', '') for x in self.binary_images_array])
self.gt_images_array = sorted(glob.glob(str(image_path) + "/epoch_30/*gt.png"))
self.gt_images_array = sorted([x.replace('gt.png', '') for x in self.gt_images_array])
self.orig_images_array = sorted(glob.glob(str(image_path) + "/epoch_30/*org.png"))
self.orig_images_array = sorted([x.replace('org.png', '') for x in self.orig_images_array])
self.in_size, self.out_size = in_size, out_size
self.data_len = len(self.binary_images_array)
def __getitem__(self, index):
single_lh_image = self.lh_images_array[index]
single_binary_image = self.binary_images_array[index]
single_gt_image = self.gt_images_array[index]
single_origin_image = self.orig_images_array[index]
lh_as_img = Image.open(single_lh_image + 'lh.png')
binary_as_img = Image.open(single_binary_image + '.png')
gt_as_img = Image.open(single_gt_image + 'gt.png')
origin_as_img = Image.open(single_origin_image + 'org.png')
lh_as_np = np.asarray(lh_as_img) / 255 # normalization?
binary_as_np = np.asarray(binary_as_img) / 255
gt_as_np = np.asarray(gt_as_img) / 255
org_as_np = np.asarray(origin_as_img) / 255
lh_as_tensor = torch.from_numpy(lh_as_np).float()
binary_as_tensor = torch.from_numpy(binary_as_np).long()
gt_as_tensor = torch.from_numpy(gt_as_np).long()
org_as_tensor = torch.from_numpy(org_as_np).float()
return org_as_tensor, binary_as_tensor, gt_as_tensor
def __len__(self):
return self.data_len
if __name__ == "__main__":
x = random.sample(range(0, 4), 3)
z = [0, 1, 2, 3]
print(z[-2:])
ss
# print(set(z) - set(x))
# ss
y = np.array([22, 33,44,55])
# print(y[x])
train = CREMIDataPreTrained('history/UNET/result_images3')
for i in range(26):
train.__getitem__(i)