-
Notifications
You must be signed in to change notification settings - Fork 0
/
datasets.py
343 lines (267 loc) · 10.3 KB
/
datasets.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
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
# --------------------------------------------------------
# Based on BEiT, timm, DINO and DeiT code bases
# https://github.com/microsoft/unilm/tree/master/beit
# https://github.com/rwightman/pytorch-image-models/tree/master/timm
# https://github.com/facebookresearch/deit
# https://github.com/facebookresearch/dino
# --------------------------------------------------------'
import os
import torch
from torchvision import datasets, transforms
from timm.data.constants import \
IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD, IMAGENET_INCEPTION_MEAN, IMAGENET_INCEPTION_STD
from imblearn.over_sampling import SMOTE
from timm.data import create_transform
from masking_generator import RandomMaskingGenerator, RandomMaskingGenerator1d
from dataset_folder import ImageFolder
import pickle
from torch.utils.data import TensorDataset
from torch.utils.data import random_split
from torch.utils.data import Dataset, DataLoader
import numpy as np
from utils import maxmin1
# class DataAugmentationForMAE(object):
# def __init__(self, args):
# imagenet_default_mean_and_std = args.imagenet_default_mean_and_std
# mean = IMAGENET_INCEPTION_MEAN if not imagenet_default_mean_and_std else IMAGENET_DEFAULT_MEAN
# std = IMAGENET_INCEPTION_STD if not imagenet_default_mean_and_std else IMAGENET_DEFAULT_STD
# self.transform = transforms.Compose([
# transforms.RandomResizedCrop(args.input_size),
# 0-1
# transforms.ToTensor(),
# transforms.Normalize(
# mean=torch.tensor(mean),
# std=torch.tensor(std))
# ])
#
# def __call__(self, image):
# self.masked_position_generator = RandomMaskingGenerator(
# args.window_size, args.mask_ratio
# )
# return self.transform(image), self.masked_position_generator()
#
# def __repr__(self):
# repr = "(DataAugmentationForBEiT,\n"
# repr += " transform = %s,\n" % str(self.transform)
# repr += " Masked position generator = %s,\n" % str(self.masked_position_generator)
# repr += ")"
# return repr
#
def DataAugmentationForMAE_1d(RG_GRFf_file_path, RG_GRFl_file_path):
#
val_percent = 0.3
# mask_ratio = 0.1 # ablation
# mask_ratio = 0.2 # ablation
# mask_ratio = 0.3 # ablation
# mask_ratio = 0.4 # ablation
# mask_ratio = 0.5 # ablation
# mask_ratio = 0.6 # ablation
# mask_ratio = 0.75 # default ablation
# mask_ratio = 0.8 # ablation
# mask_ratio = 0.9 # ablation
print ('mask_ratio:', mask_ratio)
with open(RG_GRFf_file_path, 'rb') as file:
GRFf = pickle.load(file).astype(np.float32)
file.close()
# with open(RG_GRFl_file_path, 'rb') as file:
# GRFl = pickle.load(file).astype(np.float32)
# file.close()
#
# GRFl = torch.from_numpy(GRFl)
#
# GRFl = torch.from_numpy(GRFl)
GRFf = maxmin1(GRFf)
sample_size = GRFf.shape[0]
L_size = GRFf.shape[2]
mask_samples_T = RandomMaskingGenerator1d(sample_size, L_size, mask_ratio).astype(np.float32)
print ('8***:', GRFf.dtype)
print ('8****:', mask_samples_T.dtype)
GRFf = torch.from_numpy(GRFf)
GRfm = torch.from_numpy(mask_samples_T)
# print ('GRFf:', GRFf.shape)
# print ('GRfm:', GRfm.shape)
GRFfdataset = TensorDataset(GRFf, GRfm)
# n_val = int(len(GRFfdataset) * val_percent)
# n_train = len(GRFfdataset) - n_val
# train_set, val_set = random_split(GRFfdataset, [n_train, n_val], generator=torch.Generator().manual_seed(0))
n_train = GRFfdataset
#
# window_size = 14 × 14
# mask_ratio = 0.75, 147
return n_train
def DataAugmentationForMAE_yz(RG_GRFf_file_path, RG_GRFl_file_path):
#
# val_percent = 0.3
mask_ratio = 0.75
#
with open(RG_GRFf_file_path, 'rb') as file:
GRFf = pickle.load(file).astype(np.float32)
file.close()
#
GRFf = maxmin1(GRFf)
sample_size = GRFf.shape[0]
L_size = GRFf.shape[2]
#
mask_samples_T = RandomMaskingGenerator1d(sample_size, L_size, mask_ratio).astype(np.float32)
#
yz_indices = np.random.choice(GRFf.shape[0], size = 1, replace=False)
# 1 × 10 × 101
grf_yz = GRFf[yz_indices]
# 1 × 101
bool_masked_pos_yz = mask_samples_T[yz_indices]
grf_yz = torch.from_numpy(grf_yz)
# mask
bool_masked_pos_yz = torch.from_numpy(bool_masked_pos_yz)
print ('grf_yz:', grf_yz.shape)
print ('bool_masked_pos_yz:', bool_masked_pos_yz.shape)
return grf_yz, bool_masked_pos_yz
def build_pretraining_dataset(args):
RG_GRFf_file_path = '/home/liullhappy/imageNet/rgDatasets/GRFf.pkl'
RG_GRFl_file_path = '/home/liullhappy/imageNet/rgDatasets/GRFl.pkl'
n_train = DataAugmentationForMAE_1d(RG_GRFf_file_path, RG_GRFl_file_path)
# transform = DataAugmentationForMAE(args)
# print("Data Aug = %s" % str(transform))
#
# return ImageFolder(args.data_path, transform=transform)
return n_train
def build_yz_dataset(args):
RG_GRFf_file_path = '/home/liullhappy/imageNet/rgDatasets/GRFf.pkl'
RG_GRFl_file_path = '/home/liullhappy/imageNet/rgDatasets/GRFl.pkl'
#
##
###
grf_yz, bool_masked_pos_yz = DataAugmentationForMAE_yz(RG_GRFf_file_path, RG_GRFl_file_path)
return grf_yz, bool_masked_pos_yz
def partition(RG_GRFf_file_path, RG_GRFl_file_path, abn_ratio):
with open(RG_GRFf_file_path, 'rb') as file:
GRFf = pickle.load(file).astype(np.float32)
file.close()
with open(RG_GRFl_file_path, 'rb') as file:
GRFl = pickle.load(file).astype(np.float32)
file.close()
# stat indice for 0(abnormal)/1(healthy)
GRFl_0_indice = np.where(GRFl==0)
GRFl_1_indice = np.where(GRFl==1)
# print ('6:', GRFl_0_indice)
# print ('7:', GRFl_1_indice)
# select the corresponding features for indice
GRFl_features_0 = GRFf[list(GRFl_0_indice[0])]
GRFl_features_1 = GRFf[list(GRFl_1_indice[0])]
# select the corresponding features for indice
GRFl_0 = GRFl[list(GRFl_0_indice[0])]
GRFl_1 = GRFl[list(GRFl_1_indice[0])]
# print (GRFl_features_0.shape)
# print (GRFl_features_1.shape)
# 67977 abnoraml / 16574 healthy,
# print (round(len(GRFl_0)/len(GRFl_1),3))
# print (GRFl_0.shape)
# print (GRFl_1.shape)
#
total_sample_size = GRFl_features_0.shape[0]
total_sample_range = list(np.arange(0, total_sample_size))
train_sample_size = int(GRFl_features_0.shape[0] * abn_ratio)
#
train_indices = np.random.choice(GRFl_features_0.shape[0], size = train_sample_size, replace=False)
# print ('1:', indices)
# print ('2:', type(indices))
# print ('3:', indices.shape)
GRFl_features_0_portion_train = GRFl_features_0[train_indices]
GRFl_0_portion_train = GRFl_0[train_indices]
# print ('4:', GRFl_features_0_portion)
# print ('5:', type(GRFl_features_0_portion))
# print ('6:', GRFl_features_0_portion.shape)
# print ('44:', GRFl_0_portion)
# print ('7:', type(GRFl_0_portion))
# print ('8:', GRFl_0_portion.shape)
#### Concatenate of 0(abnormal) + 1(healthy)
## 3398 + 16574 = 19972
GRFf_train = np.concatenate((GRFl_features_0_portion_train, GRFl_features_1),axis = 0)
GRFl_train = np.concatenate((GRFl_0_portion_train, GRFl_1),axis = 0)
# print ('9:', type(GRFf))
# print ('10:', GRFf.shape)
# print ('11:', type(GRFl))
# print ('12:', GRFl.shape)
#
test_indices = []
for e in total_sample_range:
if e not in train_indices:
# test_indices.append(np.where(total_sample_range==e)[0][0])
test_indices.append(e)
##
test_indices = np.array(test_indices)
GRFl_features_0_portion_test = GRFl_features_0[test_indices]
GRFl_0_portion_test = GRFl_0[test_indices]
##
GRFf_test = np.concatenate((GRFl_features_0_portion_test, GRFl_features_1), axis = 0)
GRFl_test = np.concatenate((GRFl_0_portion_test, GRFl_1), axis = 0)
####
print ('13:', GRFf_train.shape)
print ('14:', GRFl_train.shape)
print ('15:', GRFf_test.shape)
print ('16:', GRFl_test.shape)
return GRFf_train, GRFl_train, GRFf_test, GRFl_test
def balance_01(X, Y):
GRFf = X
GRFl = Y
#
smo = SMOTE(n_jobs=-1)
#
####
GRFf_1d = GRFf.reshape(GRFf.shape[0], -1)
# print ('test1:', GRFf.shape)
#
GRFf_1d_fnum = GRFf.shape[1]
# print ('1:', GRFf_1d.shape)
GRFf_re, GRFl_re = smo.fit_resample(GRFf_1d, GRFl)
# print ('2:', GRFf_re.shape)
# print ('3:', GRFl_re.shape)
#
GRFf_re = GRFf_re.reshape(GRFf_re.shape[0], GRFf_1d_fnum, -1)
#
# print ('4:', GRFf_re.shape)
#### test the distribution proportion
#
#
GRFl_re_0_indice = np.where(GRFl_re==0)
GRFl_re_1_indice = np.where(GRFl_re==1)
####
# print ('5:', len(list(GRFl_re_0_indice[0])))
# print ('6:', len(list(GRFl_re_1_indice[0])))
# print ('7:', GRFf_re)
#
#
#
print ('17:', len(list(GRFl_re_0_indice[0])))
print ('18:', len(list(GRFl_re_1_indice[0])))
return GRFf_re, GRFl_re
#
#
#
def build_dataset(args):
RG_GRFf_file_path = '/home/liullhappy/imageNet/rgDatasets/GRFf.pkl'
RG_GRFl_file_path = '/home/liullhappy/imageNet/rgDatasets/GRFl.pkl'
#
abn_ratio = 0.3 # 30%
# abn_ratio = 0.2 # 20%
# abn_ratio = 0.1 # 10%
# abn_ratio = 0.05 # 5%
# abn_ratio = 0.01 # 1%
#
#
GRFf_train_part, GRFl_train_part, GRFf_val_part, GRFl_val_part = partition(RG_GRFf_file_path, RG_GRFl_file_path, abn_ratio)
#
GRFf_train_ba, GRFl_train_ba = balance_01(GRFf_train_part, GRFl_train_part)
#
GRFf_val_ba, GRFl_val_ba = balance_01(GRFf_val_part, GRFl_val_part)
# ------------------------ 2、normlization max-min scaler------------------------
GRFf_train = torch.from_numpy(GRFf_train_ba)
GRFf_train = maxmin1(GRFf_train)
GRFf_val = torch.from_numpy(GRFf_val_ba)
GRFf_val = maxmin1(GRFf_val)
#
GRFl_train = torch.from_numpy(GRFl_train_ba)
GRFl_val = torch.from_numpy(GRFl_val_ba)
# ------------------------ 3、Tensor Dataset------------------------
n_train = TensorDataset(GRFf_train, GRFl_train)
n_test = TensorDataset(GRFf_val, GRFl_val)
return n_train, n_test, 1