-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathpreprocess_data.py
274 lines (241 loc) · 13.1 KB
/
preprocess_data.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
import os
import shutil
from PIL import Image, ImageOps, ImageFilter
import numpy as np
import matplotlib.pyplot as plt
class FixData():
def __init__(self, filepath):
self.filepath = filepath
self.train_path = self.filepath + "/train"
self.train_label_path = self.filepath + "/train_label"
self.test_path, self.test_label_path = self.rename_folders()
self.val_path = None
self.val_label_path = None
def rename_folders(self):
try:
os.rename(os.path.join(self.filepath, "val"), os.path.join(self.filepath, "test"))
os.rename(os.path.join(self.filepath, "val_label"), os.path.join(self.filepath, "test_label"))
print("Directory names have successfully been changed.")
except FileNotFoundError:
print("Folder does not exist. Either cell has already ran or folder does not exist.")
except OSError:
pass
return self.filepath + "/test", self.filepath + "/test_label"
def add_zeros(self, num):
return str(num.zfill(4))
def rename_sort_data(self):
for filename in os.listdir(self.train_path):
updated_filename = filename[44:]
updated_filename = os.path.join(self.train_path, self.add_zeros(updated_filename[:-4])+".png")
os.rename(os.path.join(self.train_path, filename), updated_filename)
for filename in os.listdir(self.train_label_path):
updated_filename = self.add_zeros(filename[44:-10])
updated_filename = os.path.join(self.train_label_path, updated_filename+".png")
os.rename(os.path.join(self.train_label_path, filename), updated_filename)
for filename in os.listdir(self.test_path):
updated_filename = self.add_zeros(filename[44:-19])
updated_filename = os.path.join(self.test_path, updated_filename+".png")
os.rename(os.path.join(self.test_path, filename), updated_filename)
for filename in os.listdir(self.test_label_path):
updated_filename = self.add_zeros(filename[44:-25])
updated_filename = os.path.join(self.test_label_path, updated_filename+".png")
os.rename(os.path.join(self.test_label_path, filename), updated_filename)
def create_val_set(self, pct):
if len(os.listdir(self.train_path)) == len(os.listdir(self.train_label_path)):
num_val_img = int(len(os.listdir(self.train_path))//(pct*100))
os.makedirs(os.path.join(self.filepath, "val"))
os.makedirs(os.path.join(self.filepath, "val_label"))
val_path = self.filepath + "/val"
val_label_path = self.filepath + "/val_label"
for img in os.listdir(self.train_path)[-num_val_img:]:
shutil.move(os.path.join(self.train_path, img), val_path + "/" + img)
for img in os.listdir(self.train_label_path)[-num_val_img:]:
shutil.move(os.path.join(self.train_label_path, img), val_label_path + "/" + img)
if len(os.listdir(val_path)) > 0 and len(os.listdir(val_label_path)) > 0:
print("Validation directories created.")
else:
return "Files in train and train_label are not the length. Please double check this."
class SplitData(FixData):
def __init__(self, filepath):
super().__init__(filepath)
self.X_train = []
self.X_val = []
self.X_test = []
self.y_train = []
self.y_val = []
self.y_test = []
self.val_path = self.filepath + "/val"
self.val_label_path = self.filepath + "/val_label"
def process_img(self, img, threshold_val, img_size, subset, rot_range):
if rot_range % 5 != 0:
raise ValueError("Number must be divisable by 5. Please input a number that is a multiple of 5")
else:
if subset == "train":
open_img = Image.open(os.path.join(self.train_path, img)).resize((img_size, img_size))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
open_img[open_img<threshold_val] = 0
open_img = Image.fromarray(open_img)
self.X_train.append(np.array(open_img)/np.max(open_img))
flip_img = ImageOps.flip(open_img)
self.X_train.append(np.array(flip_img)/np.max(flip_img))
mir_img = ImageOps.mirror(open_img)
self.X_train.append(np.array(mir_img)/np.max(mir_img))
for i in range(5,rot_range,5):
rot_img = open_img.rotate(i)
self.X_train.append(np.array(rot_img)/np.max(rot_img))
if subset == "val":
open_img = Image.open(os.path.join(self.val_path, img)).resize((img_size, img_size))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
open_img[open_img<threshold_val] = 0
open_img = Image.fromarray(open_img)
self.X_val.append(np.array(open_img)/np.max(open_img))
flip_img = ImageOps.flip(open_img)
self.X_val.append(np.array(flip_img)/np.max(flip_img))
mir_img = ImageOps.mirror(open_img)
self.X_val.append(np.array(mir_img)/np.max(mir_img))
for i in range(5,rot_range,5):
rot_img = open_img.rotate(i)
self.X_val.append(np.array(rot_img)/np.max(rot_img))
if subset == "test":
open_img = Image.open(os.path.join(self.test_path, img)).resize((img_size, img_size))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
open_img[open_img<threshold_val] = 0
open_img = Image.fromarray(open_img)
self.X_test.append(np.array(open_img)/np.max(open_img))
flip_img = ImageOps.flip(open_img)
self.X_test.append(np.array(flip_img)/np.max(flip_img))
mir_img = ImageOps.mirror(open_img)
self.X_test.append(np.array(mir_img)/np.max(mir_img))
for i in range(5,rot_range,5):
rot_img = open_img.rotate(i)
self.X_test.append(np.array(rot_img)/np.max(rot_img))
def process_mask(self, img, img_size, subset, rot_range):
if rot_range % 5 != 0:
raise ValueError("Number must be divisable by 5. Please input a number that is a multiple of 5")
else:
if subset == "train":
open_img = Image.open(os.path.join(self.train_label_path, img)).resize((img_size, img_size))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
open_img[open_img>0] = 1
open_img = Image.fromarray(open_img)
self.y_train.append(np.array(open_img))
flip_img = ImageOps.flip(open_img)
self.y_train.append(np.array(flip_img))
mir_img = ImageOps.mirror(open_img)
self.y_train.append(np.array(mir_img))
for i in range(5,rot_range,5):
rot_img = open_img.rotate(i)
self.y_train.append(np.array(rot_img))
if subset == "val":
open_img = Image.open(os.path.join(self.val_label_path, img)).resize((img_size, img_size))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
open_img[open_img>0] = 1
open_img = Image.fromarray(open_img)
self.y_val.append(np.array(open_img))
flip_img = ImageOps.flip(open_img)
self.y_val.append(np.array(flip_img))
mir_img = ImageOps.mirror(open_img)
self.y_val.append(np.array(mir_img))
for i in range(5,rot_range,5):
rot_img = open_img.rotate(i)
self.y_val.append(np.array(rot_img))
if subset == "test":
open_img = Image.open(os.path.join(self.test_label_path, img)).resize((img_size, img_size))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
open_img[open_img>0] = 1
open_img = Image.fromarray(open_img)
self.y_test.append(np.array(open_img))
flip_img = ImageOps.flip(open_img)
self.y_test.append(np.array(flip_img))
mir_img = ImageOps.mirror(open_img)
self.y_test.append(np.array(mir_img))
for i in range(5,rot_range,5):
rot_img = open_img.rotate(i)
self.y_test.append(np.array(rot_img))
def resize_img(self, img_size, threshold_val, rot_range):
for img in os.listdir(self.train_path):
self.process_img(img,threshold_val,img_size,"train",rot_range)
for img in os.listdir(self.val_path):
self.process_img(img,threshold_val,img_size,"val",rot_range)
for img in os.listdir(self.test_path):
self.process_img(img,threshold_val,img_size,"test",rot_range)
for img in os.listdir(self.train_label_path):
self.process_mask(img,img_size,"train",rot_range)
for img in os.listdir(self.val_label_path):
self.process_mask(img,img_size,"val",rot_range)
for img in os.listdir(self.test_label_path):
self.process_mask(img,img_size,"test",rot_range)
def print_img_pair(self, num):
fig = plt.figure(figsize=(10,10))
plt.subplot(2, 2, 1)
plt.imshow(np.array(self.X_train[num]), cmap="gray")
plt.subplot(2, 2, 2)
plt.imshow(np.array(self.y_train[num]), cmap="gray")
def get_shapes(self):
return f"X_train: {np.array(self.X_train).shape}, X_val: {np.array(self.X_val).shape}, X_test: {np.array(self.X_test).shape}, y_train: {np.array(self.y_train).shape}, y_val: {np.array(self.y_val).shape}, y_test: {np.array(self.y_test).shape}"
def data(self):
updated_X_train = np.array(self.X_train).reshape(int(np.array(self.X_train).ravel().shape[0])//128//128,128,128,1).astype('float64')
updated_y_train = np.array(self.y_train).reshape(int(np.array(self.y_train).ravel().shape[0])//128//128,128,128,1).astype('float64')
updated_X_val = np.array(self.X_val).reshape(int(np.array(self.X_val).ravel().shape[0])//128//128,128,128,1).astype('float64')
updated_y_val = np.array(self.y_val).reshape(int(np.array(self.y_val).ravel().shape[0])//128//128,128,128,1).astype('float64')
updated_X_test = np.array(self.X_test).reshape(int(np.array(self.X_test).ravel().shape[0])//128//128,128,128,1).astype('float64')
updated_y_test = np.array(self.y_test).reshape(int(np.array(self.y_test).ravel().shape[0])//128//128,128,128,1).astype('float64')
return updated_X_train, updated_y_train, updated_X_val, updated_y_val, updated_X_test, updated_y_test
def non_aug_data(subset,data_path):
if subset == "X_train":
aug_list = []
for img in os.listdir(data_path):
open_img = Image.open(os.path.join(data_path, img)).resize((128, 128))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
aug_list.append(open_img)
return np.array(aug_list)/np.max(aug_list)
if subset == "X_val":
aug_list = []
for img in os.listdir(data_path):
open_img = Image.open(os.path.join(data_path, img)).resize((128, 128))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
aug_list.append(open_img)
return np.array(aug_list)/np.max(aug_list)
if subset == "X_test":
aug_list = []
for img in os.listdir(data_path):
open_img = Image.open(os.path.join(data_path, img)).resize((128, 128))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
aug_list.append(open_img)
return np.array(aug_list)/np.max(aug_list)
if subset == "y_train":
aug_list = []
for img in os.listdir(data_path):
open_img = Image.open(os.path.join(data_path, img)).resize((128, 128))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
open_img[open_img>0] = 1
aug_list.append(open_img)
return np.array(aug_list)
if subset == "y_val":
aug_list = []
for img in os.listdir(data_path):
open_img = Image.open(os.path.join(data_path, img)).resize((128, 128))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
open_img[open_img>0] = 1
aug_list.append(open_img)
return np.array(aug_list)
if subset == "y_test":
aug_list = []
for img in os.listdir(data_path):
open_img = Image.open(os.path.join(data_path, img)).resize((128, 128))
open_img = ImageOps.grayscale(open_img)
open_img = np.array(open_img)
open_img[open_img>0] = 1
aug_list.append(open_img)
return np.array(aug_list)