-
Notifications
You must be signed in to change notification settings - Fork 0
/
3D_unet_data_generator.py
511 lines (342 loc) · 18 KB
/
3D_unet_data_generator.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
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
import tensorflow as tf
import keras
print(tf.__version__)
print(keras.__version__)
#Make sure the GPU is available.
from tensorflow.keras.utils import Sequence
device_name = tf.test.gpu_device_name()
if device_name != '/device:GPU:0':
raise SystemError('GPU device not found')
print('Found GPU at: {}'.format(device_name))
'''
conda remove -n spt16 --all
conda deactivate
conda deactivate
cd stelar_3d/
virtualenv spt20
source spt19/bin/activate
pip install tensorflow
pip install classification-models-3D
pip install efficientnet-3D
pip install segmentation-models-3D
pip install scikit-learn
pip install matplotlib
pip install patchify
pip install scikit-image
python 3D_unet_1.py
pip uninstall tensorflow
pip uninstall classification-models-3D
pip uninstall efficientnet-3D
pip uninstall segmentation-models-3D
pip uninstall scikit-learn
pip uninstall matplotlib
pip uninstall patchify
pip uninstall scikit-image
'''
'''
export CUDA_VISIBLE_DEVICES=0
conda deactivate
conda deactivate
cd stelar_3dunet/
source spt19/bin/activate
python3 3D_unet_continual_model_tuning_data_generator.py --crop_1 1 --crop_2 2 --crop_3 3
python3 3D_unet_continual_model_tuning_data_generator.py --crop_1 37 --crop_2 40 --crop_3 41
, ,
export CUDA_VISIBLE_DEVICES=1
conda deactivate
conda deactivate
cd stelar_3dunet/
source spt19/bin/activate
python3 3D_unet_continual_model_tuning_data_generator.py --crop_1 4 --crop_2 5 --crop_3 7
python3 3D_unet_continual_model_tuning_data_generator.py --crop_1 33 --crop_2 34 --crop_3 35
python3 3D_unet_continual_model_tuning_data_generator.py --crop_1 36 --crop_2 37 --crop_3 40
python3 3D_unet_continual_model_tuning_data_generator.py --crop_1 37 --crop_2 40 --crop_3 41
****** memory error issues *************
export CUDA_VISIBLE_DEVICES=1
conda deactivate
conda deactivate
cd stelar_3dunet/
source spt19/bin/activate
export TF_GPU_ALLOCATOR=cuda_malloc_async
python3 3D_unet_continual_model_tuning.py --crop_1 1 --crop_2 2 --crop_3 3
'''
import segmentation_models_3D as sm
from skimage import io
from patchify import patchify, unpatchify
import numpy as np
from matplotlib import pyplot as plt
from keras import backend as K
from keras.utils import to_categorical
from sklearn.model_selection import train_test_split
'''physical_devices = tf.config.list_physical_devices('GPU')
for device in physical_devices:
tf.config.experimental.set_memory_growth(device, True)'''
class DataGenerator(Sequence):
def __init__(self, X, y, batch_size=8, shuffle=True):
self.X = X
self.y = y
self.batch_size = batch_size
self.shuffle = shuffle
self.on_epoch_end()
def __len__(self):
# Number of batches per epoch
return int(np.floor(len(self.X) / self.batch_size))
def __getitem__(self, index):
# Generate one batch of data
batch_X = self.X[index * self.batch_size:(index + 1) * self.batch_size]
batch_y = self.y[index * self.batch_size:(index + 1) * self.batch_size]
return batch_X, batch_y
def on_epoch_end(self):
# Shuffle the data after each epoch if specified
if self.shuffle:
indices = np.arange(len(self.X))
np.random.shuffle(indices)
self.X = self.X[indices]
self.y = self.y[indices]
num_epochs = 600
vista_crop_dict = { 0:'NA' , 1: 'ALFALFA', 2: 'BEET', 3: 'CLOVER', 4: 'FLAX', 5: 'FLOWERING_LEGUMES', 6: 'FLOWERS', 7: 'FOREST', 8: 'GRAIN_MAIZE', 9: 'GRASSLAND', 10: 'HOPS', 11: 'LEGUMES', 12: 'VISTA_NA', 13: 'PERMANENT_PLANTATIONS', 14: 'PLASTIC', 15: 'POTATO', 16: 'PUMPKIN', 17: 'RICE', 18: 'SILAGE_MAIZE', 19: 'SOY', 20: 'SPRING_BARLEY', 21: 'SPRING_OAT', 22: 'SPRING_OTHER_CEREALS', 23: 'SPRING_RAPESEED', 24: 'SPRING_RYE', 25: 'SPRING_SORGHUM', 26: 'SPRING_SPELT', 27: 'SPRING_TRITICALE', 28: 'SPRING_WHEAT', 29: 'SUGARBEET', 30: 'SUNFLOWER', 31: 'SWEET_POTATOES', 32: 'TEMPORARY_GRASSLAND', 33: 'WINTER_BARLEY', 34: 'WINTER_OAT', 35: 'WINTER_OTHER_CEREALS', 36: 'WINTER_RAPESEED', 37: 'WINTER_RYE', 38: 'WINTER_SORGHUM', 39: 'WINTER_SPELT', 40: 'WINTER_TRITICALE', 41: 'WINTER_WHEAT'}
import argparse
parser = argparse.ArgumentParser(description='Enter crop type numbers in order')
parser.add_argument('--crop_1', type=int, default=1, help='Select crop type')
parser.add_argument('--crop_2', type=int, default=2, help='Select crop type')
parser.add_argument('--crop_3', type=int, default=3, help='Select crop type')
args = parser.parse_args()
cr_1 = args.crop_1
cr_2 = args.crop_2
cr_3 = args.crop_3
chosen_crop_types_list = [cr_1, cr_2, cr_3]
chosen_crop_types_list1 = [cr_1, cr_2, cr_3, 4, 5]
crop_types_all_list = [ 1, 2, 3, 4, 5, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 18, 19, 20, 21, 23, 27, 28, 30, 32, 33, 34, 35, 36, 37, 40, 41]
print("vista_crop_dict[chosen_crop_types_list[1]], vista_crop_dict[chosen_crop_types_list[1]], vista_crop_dict[chosen_crop_types_list[1]]", vista_crop_dict[chosen_crop_types_list[0]], vista_crop_dict[chosen_crop_types_list[1]], vista_crop_dict[chosen_crop_types_list[2]])
sampling_group_fractions = [1.0, 1.0, 1.0]
#all_input_img = []
#all_input_mask = []
all_input_img_f = []
all_input_mask_f = []
counted = 0
for crop_no in chosen_crop_types_list:
chosen_crop_type = vista_crop_dict[crop_no]
print("chosen_crop_type", chosen_crop_type)
input_img = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train_'+vista_crop_dict[crop_no]+'.tif')
input_mask = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab_'+vista_crop_dict[crop_no]+'.tif').astype(np.uint8)
print("input_img.shape", input_img.shape)
print("input_mask.shape", input_mask.shape)
input_img0 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n0.tif')
input_mask0 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n0.tif').astype(np.uint8)
print("input_img0.shape", input_img0.shape)
print("input_mask0.shape", input_mask0.shape)
input_img1 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n1.tif')
input_mask1 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n1.tif').astype(np.uint8)
print("input_img1.shape", input_img1.shape)
print("input_mask1.shape", input_mask1.shape)
input_img2 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n2.tif')
input_mask2 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n2.tif').astype(np.uint8)
print("input_img2.shape", input_img2.shape)
print("input_mask2.shape", input_mask2.shape)
input_img3 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n3.tif')
input_mask3 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n3.tif').astype(np.uint8)
print("input_img3.shape", input_img3.shape)
print("input_mask3.shape", input_mask3.shape)
input_img4 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n4.tif')
input_mask4 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n4.tif').astype(np.uint8)
print("input_img4.shape", input_img4.shape)
print("input_mask4.shape", input_mask4.shape)
input_img5 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n5.tif')
input_mask5 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n5.tif').astype(np.uint8)
print("input_img5.shape", input_img5.shape)
print("input_mask5.shape", input_mask5.shape)
input_img6 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n6.tif')
input_mask6 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n6.tif').astype(np.uint8)
print("input_img6.shape", input_img6.shape)
print("input_mask6.shape", input_mask6.shape)
input_img7 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n7.tif')
input_mask7 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n7.tif').astype(np.uint8)
print("input_img7.shape", input_img7.shape)
print("input_mask7.shape", input_mask7.shape)
input_img8 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n8.tif')
input_mask8 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n8.tif').astype(np.uint8)
print("input_img8.shape", input_img8.shape)
print("input_mask8.shape", input_mask8.shape)
input_img9 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n9.tif')
input_mask9 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n9.tif').astype(np.uint8)
print("input_img9.shape", input_img9.shape)
print("input_mask9.shape", input_mask9.shape)
input_img10 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/train'+vista_crop_dict[crop_no]+'n10.tif')
input_mask10 = io.imread('/home/luser/stelar_3dunet/storage/per_crop_data_labels/'+vista_crop_dict[crop_no]+'/lab'+vista_crop_dict[crop_no]+'n10.tif').astype(np.uint8)
print("input_img9.shape", input_img9.shape)
print("input_mask9.shape", input_mask9.shape)
input_img_f = np.concatenate((input_img, input_img0, input_img1, input_img2, input_img3, input_img4, input_img5, input_img6, input_img7, input_img8, input_img9, input_img10), axis=0)
input_mask_f = np.concatenate((input_mask, input_mask0, input_mask1, input_mask2, input_mask3, input_mask4, input_mask5, input_mask6, input_mask7, input_mask8, input_mask9, input_mask10), axis=0)
bis = int(len(input_img_f)*sampling_group_fractions[counted]) - 2
print("bis", bis)
print("before input_img_f.shape", input_img_f.shape)
print("before input_mask_f.shape", input_mask_f.shape)
input_img_f = input_img_f[:bis]
input_mask_f = input_mask_f[:bis]
print("after input_img_f.shape", input_img_f.shape)
print("after input_mask_f.shape", input_mask_f.shape)
all_input_img_f.append(input_img_f)
all_input_mask_f.append(input_mask_f)
counted+=1
all_input_img_f = np.concatenate((all_input_img_f), axis=0)
all_input_mask_f = np.concatenate((all_input_mask_f), axis=0)
print("all_input_img_f.shape", all_input_img_f.shape)
print("all_input_mask_f.shape", all_input_mask_f.shape)
input_img = np.concatenate((all_input_img_f), axis=0).reshape(-1, 64, 64, 64)
#input_img = np.array(all_input_img).reshape(-1, 64, 64, 64)
#input_mask = np.array(all_input_mask).reshape(-1, 64, 64)
input_mask = np.concatenate((all_input_mask_f), axis=0).reshape(-1, 64, 64)
del all_input_img_f
del all_input_mask_f
del input_img_f
del input_mask_f
#unique_elements, element_counts = np.unique(input_mask, return_counts=True)
#print("1 : unique_elements, element_counts", unique_elements, element_counts)
#selected_counts = np.array([element_counts[1], element_counts[2], element_counts[3]])
input_mask = np.repeat(input_mask[:, np.newaxis, :, :], repeats=64, axis=1)
#np.random.shuffle(input_mask)
#np.random.shuffle(input_img)
#input_img = input_img[:10000]
#input_mask = input_mask[:10000]
print("final input_img.shape", input_img.shape)
print("final input_mask.shape", input_mask.shape)
#all_input_img = 0
#all_input_mask = 0
input_mask[input_mask<chosen_crop_types_list[0]]=0
input_mask[input_mask>chosen_crop_types_list[-1]]=0
for i in range(3):
input_mask[input_mask==chosen_crop_types_list[i]]=i+1
###########################################################################################################
unique_elements, element_counts = np.unique(input_mask, return_counts=True)
print("2 : unique_elements, element_counts", unique_elements, element_counts)
# selected_counts = np.array([element_counts[0], element_counts[np.where(unique_elements == chosen_crop_types_list[0])[0][0]], element_counts[np.where(unique_elements == chosen_crop_types_list[1])[0][0]], element_counts[np.where(unique_elements == chosen_crop_types_list[2])[0][0]]])
#selected_counts = np.array([element_counts[0], element_counts[np.where(unique_elements == 1)[0][0]], element_counts[np.where(unique_elements == 2)[0][0]], element_counts[np.where(unique_elements == 3)[0][0]]])
selected_counts = np.array(element_counts)
selected_counts_fractions = selected_counts/np.sum(selected_counts)
selected_counts_fractions_f = selected_counts_fractions**(-1)
weights = selected_counts_fractions_f/np.sum(selected_counts_fractions_f)
del selected_counts_fractions_f
del selected_counts_fractions
del selected_counts
del element_counts
del unique_elements
print("weights", weights)
print("weights[0]", weights[0])
print("weights[1]", weights[1])
print("weights[2]", weights[2])
print("weights[3]", weights[3])
###########################################################################################################
#input_mask = np.repeat(input_mask[:, np.newaxis, :, :], repeats=64, axis=1)
#input_img[input_img==0] = np.median(input_img)
'''for m in range(64):
for i in range(64):
for j in range(64):
input_img[m,:,i,j][input_img[m,:,i,j]==0] = np.median(input_img[m,:,i,j]).astype(np.uint8)'''
lai_uniques = np.unique(input_img)
for n in range(len(lai_uniques)):
input_img[input_img==lai_uniques[n]]=n
input_img = input_img.astype(np.uint8)
lai_uniques = 0
n_classes=4
train_img = np.stack((input_img,)*3, axis=-1)
train_mask = np.expand_dims(input_mask, axis=4)
train_mask_cat = to_categorical(train_mask, num_classes=n_classes)
X_train, X_test, y_train, y_test = train_test_split(train_img, train_mask_cat, test_size = 0.10, random_state = 0)
'''X_train = X_train[:9000]
y_train = y_train[:9000]
X_test = X_test[:1000]
y_test = y_test[:1000]'''
print("X_train.shape", X_train.shape)
print("X_test.shape", X_test.shape)
print("y_train.shape", y_train.shape)
print("y_test.shape", y_test.shape)
del train_mask_cat
del train_mask
del train_img
del input_img
del input_mask
del input_img0
del input_mask0
del input_img1
del input_mask1
del input_img2
del input_mask2
del input_img3
del input_mask3
del input_img4
del input_mask4
del input_img5
del input_mask5
'''train_img = 0
train_mask_cat = 0
train_mask = 0
input_mask = 0
input_img = 0'''
def dice_coefficient(y_true, y_pred):
smoothing_factor = 1
flat_y_true = K.flatten(y_true)
flat_y_pred = K.flatten(y_pred)
return (2. * K.sum(flat_y_true * flat_y_pred) + smoothing_factor) / (K.sum(flat_y_true) + K.sum(flat_y_pred) + smoothing_factor)
def dice_coefficient_loss(y_true, y_pred):
return 1 - dice_coefficient(y_true, y_pred)
encoder_weights = 'imagenet'
BACKBONE = 'vgg16' #Try vgg16, efficientnetb7, inceptionv3, resnet50
activation = 'softmax'
patch_size = 64
n_classes = 4
channels=3
LR = 0.0001
#LR = 0.000001
optim = keras.optimizers.Adam(LR)
dice_loss = sm.losses.DiceLoss(class_weights=np.array([weights[0], weights[1], weights[2], weights[3]]))
focal_loss = sm.losses.CategoricalFocalLoss()
total_loss = dice_loss + (1 * focal_loss)
metrics = [sm.metrics.IOUScore(threshold=0.5), sm.metrics.FScore(threshold=0.5)]
preprocess_input = sm.get_preprocessing(BACKBONE)
X_train_prep = preprocess_input(X_train)
X_test_prep = preprocess_input(X_test)
del X_train
del X_test
#Define the model. Here we use Unet but we can also use other model architectures from the library.
model = sm.Unet(BACKBONE, classes=n_classes,
input_shape=(patch_size, patch_size, patch_size, channels),
encoder_weights=encoder_weights,
activation=activation)
model.compile(optimizer = optim, loss=total_loss, metrics=metrics)
print(model.summary())
train_generator = DataGenerator(X_train_prep, y_train, batch_size=8)
validation_generator = DataGenerator(X_test_prep, y_test, batch_size=8)
history = model.fit(train_generator,
epochs=num_epochs,
verbose=1,
validation_data=validation_generator)
'''X_train_prep_c = X_train_prep[7000:]
y_train_c = y_train[7000:]
history=model.fit(X_train_prep_c,
y_train_c,
batch_size=8,
epochs=num_epochs,
verbose=1,
validation_data=(X_test_prep, y_test))
del X_train_prep_c
del y_train_c
X_train_prep_c = X_train_prep[:9000]
y_train_c = y_train[:9000]
history=model.fit(X_train_prep_c,
y_train_c,
batch_size=8,
epochs=num_epochs,
verbose=1,
validation_data=(X_test_prep, y_test))'''
model.save('/home/luser/stelar_3dunet/storage/data_gen_model/3D_unet_res_labels_'+vista_crop_dict[chosen_crop_types_list[0]]+'_'+vista_crop_dict[chosen_crop_types_list[1]]+'_'+vista_crop_dict[chosen_crop_types_list[2]]+'_num_epocs_'+str(num_epochs)+'.h5')
'''for i in range(X_train_prep.shape[0]//2000):
X_train_prep_c = X_train_prep[i*2000:(i+1)*2000]
y_train_c = y_train[i*2000:(i+1)*2000]
history=model.fit(X_train_prep_c,
y_train_c,
batch_size=8,
epochs=num_epochs,
verbose=1,
validation_data=(X_test_prep, y_test))
X_train_prep_c = 0
y_train_c = 0'''