-
Notifications
You must be signed in to change notification settings - Fork 6
/
load_dataset.py
280 lines (249 loc) · 12.2 KB
/
load_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
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
# -*- coding: utf-8 -*-
"""
//////////////////////////////////////////////////////////////////////////////////////////
// Original author: Aritz Lizoain
// Github: https://github.com/aritzLizoain
// My personal website: https://aritzlizoain.github.io/
// Description: CNN Image Segmentation
// Copyright 2020, Aritz Lizoain.
// License: MIT License
//////////////////////////////////////////////////////////////////////////////////////////
- load_images (unused)
- get_weights: calculates the weights for the loss function
- process_fits: loads FITS files and creates small sections
- images_small2big: reconstructs small sections
- check_one_object: looks for the chosen category section by section
"""
import os
import sys
import numpy as np
import cv2
from skimage.transform import resize
##############################################################
# NOT USED IN VERSION 2.0.
# THE IMAGES ARE NOW SAVED AND LOADED AS ARRAYS, NOT AS PNG FILES
# def load_images(TRAIN_PATH='', TEST_PATH='',\
# TEST_PREDICTIONS_PATH='',IMG_WIDTH = \
# 256, IMG_HEIGHT = 256):
# train_ids = next(os.walk(TRAIN_PATH))[2]
# test_ids = next(os.walk(TEST_PATH))[2]
# # Get and resize train images and masks
# images = np.zeros((len(train_ids), IMG_HEIGHT, IMG_WIDTH,3)\
# , dtype=np.uint8)
# test_images = np.zeros((len(test_ids), IMG_HEIGHT, IMG_WIDTH\
# , 3), dtype=np.uint8)
# sys.stdout.flush()
# # # train images
# for n,id_ in enumerate(train_ids):
# img = cv2.imread(TRAIN_PATH + id_)
# img = resize(img, (IMG_HEIGHT, IMG_WIDTH), mode='constant'\
# , preserve_range=True)
# images[n] = img
# # # test images
# for n,id_ in enumerate(test_ids):
# mask_ = cv2.imread(TEST_PATH + id_)
# mask_ = resize(mask_, (IMG_HEIGHT, IMG_WIDTH),\
# preserve_range=True, mode='constant')
# test_images[n] = mask_
# print('Dataset correctly loaded')
# return images, test_images
#-------------------------------------------------------------
def get_weights(images,test_images):
from mask import get_percentages
#all_images = np.concatenate((images, test_images)) to take
#both training and test images
all_images=images #to take only training images
unique_elements, percentage = get_percentages(all_images)
inverse_percentages=1/percentage #the weights are inversely
#proportional to their frequency
weights = inverse_percentages/sum(inverse_percentages)*\
len(unique_elements) #normalize to the number of classes
return weights
#-------------------------------------------------------------
def process_fits(name='name.fits', size=256, normalized='yes'\
, normalization_value=255):
import matplotlib.pyplot as plt
from astropy.visualization import astropy_mpl_style
plt.style.use(astropy_mpl_style)
from astropy.utils.data import get_pkg_data_filename
from astropy.io import fits
import numpy as np
#LOADING THE IMAGE AND GETTING INFORMATION
image_file = get_pkg_data_filename(name)
image_data = fits.getdata(image_file, ext=0)
# image_data=image_data/100
# normalize
if normalized=='yes':
maximum_value=np.amax(image_data)
image_data_normalized=image_data/maximum_value*\
normalization_value
elif normalized=='no':
# image_data=image_data
None
else:
print(' ERROR: The given input for the normalization\
variable is not an option. Please choose yes/no')
#information about the original full image
image_length=image_data.shape[1]
image_height=image_data.shape[0]
amount_images_wide=int((image_length/2)/size) #we will only
#take half of the image
amount_images_high=int(image_height/size)
# # RESIZE image UNUSED
# if image_length/size-amount_images_wide < 0.5:
# amount_images_wide=amount_images_wide
# else:
# amount_images_wide=amount_images_wide + 1
# if image_height/size-amount_images_high < 0.5:
# amount_images_high=amount_images_high
# else:
# amount_images_high=amount_images_high + 1
# number_of_images=amount_images_wide*amount_images_high
# if normalized=='yes':
# image_data_normalized_resized=np.resize(image_data_normalized, (size*amount_images_high, size*amount_images_wide))
# print(' Resized and normalized real test image shape: {0}'.format(image_data_normalized_resized.shape))
# plt.figure()
# plt.imshow(image_data_normalized_resized)
# plt.colorbar()
# plt.title('Normalized and resized real test image', fontsize=15)
# plt.show()
# image_data_use = image_data_normalized_resized
# elif normalized=='no':
# image_data_resized=np.resize(image_data, (size*amount_images_high, size*amount_images_wide))
# print(' Resized real test image shape: {0}'.format(image_data_resized.shape))
# plt.figure()
# plt.imshow(image_data_resized)
# plt.colorbar()
# plt.title('Resized real test image', fontsize=25)
# plt.show()
# image_data_use = image_data_resized
#CUT
number_of_images = amount_images_wide*amount_images_high
image_data_use=np.zeros((amount_images_high*size,amount_images_wide*size))
starting_value=image_data.shape[1]-image_data_use.shape[1]
if normalized=='yes':
for i in range(0,image_data_use.shape[0]):
for j in range (0,image_data_use.shape[1]):
image_data_use[i,j] = image_data_normalized[i,j + starting_value]
print(' Cut and normalized real test image shape: {0}'.format(image_data_use.shape))
plt.figure()
plt.grid(False)
plt.imshow(image_data_use)
plt.colorbar()
plt.title('Normalized and cut real test image', fontsize=15)
plt.show()
elif normalized=='no':
for i in range(0,image_data_use.shape[0]):
for j in range (0,image_data_use.shape[1]):
image_data_use[i,j] = image_data[i,j + starting_value]
plt.figure()
plt.grid(False)
plt.imshow(image_data_use)
plt.colorbar()
plt.title('Cut real test image', fontsize=20)
plt.show()
print(' Cut real test image shape: {0}'.format(image_data_use.shape))
# Create the smaller sections
print(' Creating {1} sections of size {0}X{0}...'.format(size, number_of_images))
images_small=np.zeros((number_of_images,size,size))
# print(' Images small shape: {0}'.format(images_small.shape))
for i in range(0, amount_images_wide):
for j in range(0, amount_images_high):
for x in range(0, size):
for y in range (0, size):
images_small[i+j*(amount_images_wide),y,x]=image_data_use[y+j*size,x+i*size]
print(' Real test images correctly created')
details=np.array([size, amount_images_high, amount_images_wide], dtype=int)
return image_data_use, images_small, details
#----------------------------------------------------------------------------
# from mask input of (n_sections, size, size, 4) gives mask output of (size, size, 4)
def images_small2big(images, details):
# Create the big image from small sections
size = details[0]
amount_images_high = details[1]
amount_images_wide = details[2]
dimensions = images.shape[3]
full_image_empty = np.zeros((size*amount_images_high, size*amount_images_wide, dimensions))
print(' Creating the real predicted test image from the {0} sections...'.format(len(images)))
for i in range(0, amount_images_wide):
for j in range(0, amount_images_high):
for x in range(0, size):
for y in range (0, size):
full_image_empty[y+j*size,x+i*size] = images[i+j*(amount_images_wide),y,x]
print(' Real test image prediction correctly created')
return full_image_empty
#----------------------------------------------------------------------------
# CHECK THE ONES WITH A SPECIFIC OBJECT IN SMALL SECTIONS
def check_one_object(test_outputs_real, test_images_real, object_to_find='Cluster', real_percentages=[0,0,0,0], details=[0,0,0]):
from mask import get_max_in_mask, mask_to_label
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
if object_to_find=='Background':
object_number = 0
elif object_to_find=='Glowing':
object_number = 1
elif object_to_find=='Hot pixel':
object_number = 2
elif object_to_find=='Cluster':
object_number = 3
else:
print(' ERROR: The given input for the object to find variable is not an option.\
Please choose background/glowing/hot pixel/cluster')
#Legend 1
red_patch = mpatches.Patch(color=[1, 0.2, 0.2], label='Cluster')
blue_patch = mpatches.Patch(color=[0,0.5,1.], label='Hot pixel')
green_patch = mpatches.Patch(color=[0.35,1.,0.25], label='Glowing')
black_patch = mpatches.Patch(color=[0./255, 0./255, 0./255], label='Background')
counter = 0
for i in range (len(test_outputs_real)):
check=test_outputs_real[i]
check=check[np.newaxis, ...]
check=get_max_in_mask(check)
is_there=object_number in check
#in order to know the position of each section
ychange = int(i/details[2])*details[0] #y axis position
xchange = (i-int(i/details[2])*details[2])*details[0] #x axis position
if is_there == True:
from mask import output_to_label_one_object
label_with_one_object = output_to_label_one_object(check, object_number)
label_all_objects = mask_to_label(check, to_print='no')
fig, ax = plt.subplots(1, 3, figsize=(20, 10))
# plt.setp(ax, xticklabels=pixels, yticklabels=pixels)
ax[0].grid(False)
ax0 = ax[0].imshow(np.squeeze(test_images_real[i]))
ax[0].set_title('Section {0}'.format(i+1), fontsize=25);
ax[0].set_xlabel('pixels', fontsize=16)
ax[0].set_ylabel('pixels', fontsize=16)
ax[0].set_xticks([0,50,100,150,200,250])
ax[0].set_xticklabels([0+xchange,50+xchange,100+xchange,150+xchange,200+xchange,250+xchange])
ax[0].set_yticks([0,50,100,150,200,250])
ax[0].set_yticklabels([0+ychange,50+ychange,100+ychange,150+ychange,200+ychange,250+ychange])
cax = fig.add_axes([0.12, 0.16, 0.25, 0.03])
plt.colorbar(ax0, orientation="horizontal", cax=cax)
ax[1].grid(False)
ax[1].imshow(label_all_objects[0])
ax[1].set_title('Predicted label', fontsize=25);
ax[1].set_xlabel('pixels', fontsize=16)
ax[1].set_ylabel('pixels', fontsize=16)
ax[1].set_xticks([0,50,100,150,200,250])
ax[1].set_xticklabels([0+xchange,50+xchange,100+xchange,150+xchange,200+xchange,250+xchange])
ax[1].set_yticks([0,50,100,150,200,250])
ax[1].set_yticklabels([0+ychange,50+ychange,100+ychange,150+ychange,200+ychange,250+ychange])
ax[2].grid(False)
ax[2].imshow(label_with_one_object[0])
ax[2].set_title('Finding {0}'.format(object_to_find), fontsize=25);
ax[2].set_xlabel('pixels', fontsize=16)
ax[2].set_ylabel('pixels', fontsize=16)
ax[2].set_xticks([0,50,100,150,200,250])
ax[2].set_xticklabels([0+xchange,50+xchange,100+xchange,150+xchange,200+xchange,250+xchange])
ax[2].set_yticks([0,50,100,150,200,250])
ax[2].set_yticklabels([0+ychange,50+ychange,100+ychange,150+ychange,200+ychange,250+ychange])
plt.legend(loc='upper center', bbox_to_anchor=(2.1, 1.5), fontsize=16,\
handles=[red_patch, blue_patch, green_patch, black_patch], ncol=4)
plt.show() #the image is not being saved
counter=counter + 1
# print(' {1} found in section {0}'.format(i, object_to_find))
else:
counter=counter
print(' {1} found in {0} sections'.format(counter, object_to_find))
return None