-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathmini2hsi-SS-5-IP.py
1243 lines (1062 loc) · 49.5 KB
/
mini2hsi-SS-5-IP.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
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.autograd import Variable
from torch.optim.lr_scheduler import StepLR
from torch.utils.data import DataLoader, Dataset
from dataset import TensorDataset
from torch.utils.data.sampler import Sampler
import numpy as np
import os
import math
import argparse
import scipy as sp
import scipy.stats
import pickle
import random
import scipy.io as sio
from sklearn.decomposition import PCA
from sklearn import metrics
import matplotlib.pyplot as plt
from scipy.io import loadmat
from sklearn import preprocessing
from sklearn.neighbors import KNeighborsClassifier
# from matplotlib import pyplot
# import matplotlib.pyplot as plt
# from matplotlib.colors import ListedColormap
import time
import task_generator_mini as tg
from torch.nn import init
parser = argparse.ArgumentParser(description="One Shot Visual Recognition")
parser.add_argument("-f","--feature_dim",type = int, default = 160) # 32,64;32,24;32,64;64,64
parser.add_argument("-d","--tar_input_dim",type = int, default = 200) # PaviaU=103；salinas=204； IN=200; KSC=176;Botswana = 144
parser.add_argument("-w","--class_num",type = int, default = 16)
parser.add_argument("-s","--shot_num_per_class",type = int, default = 1)
parser.add_argument("-b","--query_num_per_class",type = int, default = 19) # 19 76 15 75
parser.add_argument("-e","--episode",type = int, default= 10000)
parser.add_argument("-t","--test_episode", type = int, default = 200)
parser.add_argument("-l","--learning_rate", type = float, default = 0.001)
parser.add_argument("-g","--gpu",type=int, default=0)
parser.add_argument("-u","--hidden_unit",type=int,default=10)
args = parser.parse_args(args=[])
# Hyper Parameters
FEATURE_DIM = args.feature_dim
TAR_INPUT_DIMENSION = args.tar_input_dim
CLASS_NUM = args.class_num
SHOT_NUM_PER_CLASS = args.shot_num_per_class
QUERY_NUM_PER_CLASS = args.query_num_per_class
EPISODE = args.episode
TEST_EPISODE = args.test_episode
LEARNING_RATE = args.learning_rate
GPU = args.gpu
HIDDEN_UNIT = args.hidden_unit
# ROOT = ".."
# 微调
TEST_CLASS_NUM = 16
TEST_SHOT_NUM_PER_CLASS = 1
TEST_QUERY_NUM_PER_CLASS = 4
# 自定义测试数据集
import torch.utils.data as data
class matcifar(data.Dataset):
"""`CIFAR10 <https://www.cs.toronto.edu/~kriz/cifar.html>`_ Dataset.
Args:
root (string): Root directory of dataset where directory
``cifar-10-batches-py`` exists.
train (bool, optional): If True, creates dataset from training set, otherwise
creates from test set.
transform (callable, optional): A function/transform that takes in an PIL image
and returns a transformed version. E.g, ``transforms.RandomCrop``
target_transform (callable, optional): A function/transform that takes in the
target and transforms it.
download (bool, optional): If true, downloads the dataset from the internet and
puts it in root directory. If dataset is already downloaded, it is not
downloaded again.
"""
def __init__(self, imdb, train, d, medicinal):
self.train = train # training set or test set
self.imdb = imdb
self.d = d
self.x1 = np.argwhere(self.imdb['set'] == 1)
self.x2 = np.argwhere(self.imdb['set'] == 3)
self.x1 = self.x1.flatten()
self.x2 = self.x2.flatten()
# if medicinal==4 and d==2:
# self.train_data=self.imdb['data'][self.x1,:]
# self.train_labels=self.imdb['Labels'][self.x1]
# self.test_data=self.imdb['data'][self.x2,:]
# self.test_labels=self.imdb['Labels'][self.x2]
if medicinal == 1:
self.train_data = self.imdb['data'][self.x1, :, :, :]
self.train_labels = self.imdb['Labels'][self.x1]
self.test_data = self.imdb['data'][self.x2, :, :, :]
self.test_labels = self.imdb['Labels'][self.x2]
else:
self.train_data = self.imdb['data'][:, :, :, self.x1]
self.train_labels = self.imdb['Labels'][self.x1]
self.test_data = self.imdb['data'][:, :, :, self.x2]
self.test_labels = self.imdb['Labels'][self.x2]
if self.d == 3:
self.train_data = self.train_data.transpose((3, 2, 0, 1)) ##(17, 17, 200, 10249)
self.test_data = self.test_data.transpose((3, 2, 0, 1))
else:
self.train_data = self.train_data.transpose((3, 0, 2, 1))
self.test_data = self.test_data.transpose((3, 0, 2, 1))
def __getitem__(self, index):
"""
Args:
index (int): Index
Returns:
tuple: (image, target) where target is index of the target class.
"""
if self.train:
img, target = self.train_data[index], self.train_labels[index]
else:
img, target = self.test_data[index], self.test_labels[index]
return img, target
def __len__(self):
if self.train:
return len(self.train_data)
else:
return len(self.test_data)
def same_seeds(seed):
torch.manual_seed(seed)
if torch.cuda.is_available():
torch.cuda.manual_seed(seed)
torch.cuda.manual_seed_all(seed) # if you are using multi-GPU.
np.random.seed(seed) # Numpy module.
random.seed(seed) # Python random module.
torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True
same_seeds(0)
import matplotlib.pyplot as plt
def classification_map(map, groundTruth, dpi, savePath):
fig = plt.figure(frameon=False)
fig.set_size_inches(groundTruth.shape[1]*2.0/dpi, groundTruth.shape[0]*2.0/dpi)
ax = plt.Axes(fig, [0., 0., 1., 1.])
ax.set_axis_off()
ax.xaxis.set_visible(False)
ax.yaxis.set_visible(False)
fig.add_axes(ax)
ax.imshow(map)
fig.savefig(savePath, dpi = dpi)
return 0
def mean_confidence_interval(data, confidence=0.95):
a = 1.0*np.array(data)
n = len(a)
m, se = np.mean(a), scipy.stats.sem(a)
h = se * sp.stats.t._ppf((1+confidence)/2., n-1)
return m,h
from operator import truediv
def AA_andEachClassAccuracy(confusion_matrix):
counter = confusion_matrix.shape[0]
list_diag = np.diag(confusion_matrix)
list_raw_sum = np.sum(confusion_matrix, axis=1)
each_acc = np.nan_to_num(truediv(list_diag, list_raw_sum))
average_acc = np.mean(each_acc)
return each_acc, average_acc
# DATA
def flip(data):
y_4 = np.zeros_like(data)
y_1 = y_4
y_2 = y_4
first = np.concatenate((y_1, y_2, y_1), axis=1)
second = np.concatenate((y_4, data, y_4), axis=1)
third = first
Data = np.concatenate((first, second, third), axis=0)
return Data
def load_data(image_file, label_file):
image_data = sio.loadmat(image_file)
label_data = sio.loadmat(label_file)
data_key = image_file.split('/')[-1].split('.')[0]
label_key = label_file.split('/')[-1].split('.')[0]
data_all = image_data[data_key] # dic-> narray , KSC:ndarray(512,217,204)
GroundTruth = label_data[label_key]
[nRow, nColumn, nBand] = data_all.shape
print(data_key, nRow, nColumn, nBand)
data = data_all.reshape(np.prod(data_all.shape[:2]), np.prod(data_all.shape[2:])) # (111104,204)
data_scaler = preprocessing.scale(data) #标准化 (X-X_mean)/X_std,
Data_Band_Scaler = data_scaler.reshape(data_all.shape[0], data_all.shape[1],data_all.shape[2])
return Data_Band_Scaler, GroundTruth # image:(512,217,3),label:(512,217)
def radiation_noise(data, alpha_range=(0.9, 1.1), beta=1/25):
alpha = np.random.uniform(*alpha_range) #从一个均匀分布[low,high)中随机采样
noise = np.random.normal(loc=0., scale=1.0, size=data.shape) # 随机高斯噪声
return alpha * data + beta * noise
def flip_augmentation(data): # arrays tuple 0:(7, 7, 103) 1=(7, 7)
horizontal = np.random.random() > 0.5 # True
vertical = np.random.random() > 0.5 # False
if horizontal:
data = np.fliplr(data) # 将数组左右翻转 arrays=[list] 0=(7, 7, 103) 1=(7, 7)
if vertical:
data = np.flipud(data) # 将数组上下翻转
return data
# 从band_selection到loader
def get_train_test_loader(Data_Band_Scaler, GroundTruth, class_num, shot_num_per_class):
print(Data_Band_Scaler.shape)
[nRow, nColumn, nBand] = Data_Band_Scaler.shape
'''label start'''
num_class = int(np.max(GroundTruth))
data_band_scaler = flip(Data_Band_Scaler) # (1830, 1020, 103)
groundtruth = flip(GroundTruth) # (1830, 1020)
del Data_Band_Scaler
del GroundTruth
HalfWidth = 16
G = groundtruth[nRow - HalfWidth:2 * nRow + HalfWidth, nColumn - HalfWidth:2 * nColumn + HalfWidth] # (642, 372)
data = data_band_scaler[nRow - HalfWidth:2 * nRow + HalfWidth, nColumn - HalfWidth:2 * nColumn + HalfWidth, :] # (642, 372, 103)
[Row, Column] = np.nonzero(G) # (42776,) (42776,) 根据G确定样本所在的行和列
# print(Row)
del data_band_scaler
del groundtruth
nSample = np.size(Row)
print('number of sample', nSample)
# 改成每类取5个样本
# np.random.seed(1334) # 可复现 IN= 1224
train = {}
test = {}
da_train = {}
m = int(np.max(G)) # 9
nlabeled = TEST_SHOT_NUM_PER_CLASS + TEST_QUERY_NUM_PER_CLASS
print('labeled number:', nlabeled)
print((200 - nlabeled) / nlabeled + 1) # 40.0
print(math.ceil((200 - nlabeled) / nlabeled) + 1) # 40
for i in range(m):
indices = [j for j, x in enumerate(Row.ravel().tolist()) if G[Row[j], Column[j]] == i + 1] # G ndarray Row中的索引
np.random.shuffle(indices) #
nb_val = shot_num_per_class # 4
train[i] = indices[:nb_val] #
da_train[i] = []
for j in range(math.ceil((200 - nlabeled) / nlabeled) + 1): # 25
da_train[i] += indices[:nb_val]
# da_train[i] = indices[:nb_val] + indices[:nb_val] + indices[:nb_val] + indices[:nb_val] + indices[:nb_val] #
test[i] = indices[nb_val:] # train_test_split 样本测试
train_indices = []
test_indices = []
da_train_indices = []
for i in range(m):
train_indices += train[i]
test_indices += test[i]
da_train_indices += da_train[i]
# np.random.shuffle(train_indices) # 训练样本不可以乱序
np.random.shuffle(test_indices) # 枚举测试样本，乱序，每个batch测试的样本最好是不同的类
# return train_indices, test_indices
print('the number of train_indices:', len(train_indices)) # 520
print('the number of test_indices:', len(test_indices)) # 9729
print('the number of train_indices after data argumentation:', len(da_train_indices)) # 520
# print(da_train_indices)
# total_size = nSample # 样本总数 10249
nTrain = len(train_indices) # 520
nTest = len(test_indices) # total_size - nTrain #9729
da_nTrain = len(da_train_indices) # 520
trainX = np.zeros([nTrain, nBand, 2 * HalfWidth + 1, 2 * HalfWidth + 1], dtype=np.float32)
trainY = np.zeros([nTrain], dtype=np.int64)
testX = np.zeros([nTest, nBand, 2 * HalfWidth + 1, 2 * HalfWidth + 1], dtype=np.float32)
testY = np.zeros([nTest], dtype=np.int64)
RandPerm = train_indices + test_indices
RandPerm = np.array(RandPerm)
for i in range(nTrain):
trainX[i, :, :, :] = np.transpose(data[Row[RandPerm[i]] - HalfWidth: Row[RandPerm[i]] + HalfWidth + 1, \
Column[RandPerm[i]] - HalfWidth: Column[RandPerm[i]] + HalfWidth + 1, :], (2,0,1))
trainY[i] = G[Row[RandPerm[i]], Column[RandPerm[i]]].astype(np.int64)
trainY = trainY - 1
# 测试:train_loader(45=5*9)
train_dataset = TensorDataset(data_tensor=trainX, target_tensor=trainY)
train_loader = DataLoader(train_dataset, batch_size=class_num * shot_num_per_class, shuffle=False, num_workers=0)
print("train data set is ok")
del train_dataset
for i in range(nTest):
testX[i, :, :, :] = np.transpose(data[ Row[RandPerm[i + nTrain ]] - HalfWidth: Row[RandPerm[i + nTrain]] + HalfWidth + 1, \
Column[RandPerm[i + nTrain]] - HalfWidth: Column[RandPerm[ i + nTrain ]] + HalfWidth + 1,:],(2,0,1))
testY[i] = G[Row[RandPerm[i + nTrain ]], Column[RandPerm[i + nTrain ]]].astype( np.int64)
print("test data set is ok")
testY = testY - 1
test_dataset = TensorDataset(data_tensor=testX, target_tensor=testY)
test_loader = DataLoader(test_dataset, batch_size=100, shuffle=False, num_workers=0)
print("test loader and train loader is OK")
del test_dataset
# # 原始数据
# imdb_ori_train = {}
# imdb_ori_train['data'] = imdb['data'][:,:,:, :nTrain]
# imdb_ori_train['Labels'] = imdb['Labels'][:nTrain]
# 目标域增强数据，训练
imdb_da_train = {}
imdb_da_train['data'] = np.zeros([2 * HalfWidth + 1, 2 * HalfWidth + 1, nBand, da_nTrain],
dtype=np.float32) # (33, 33, 103, 1800)
imdb_da_train['Labels'] = np.zeros([da_nTrain], dtype=np.int64)
imdb_da_train['set'] = np.zeros([da_nTrain], dtype=np.int64)
da_RandPerm = np.array(da_train_indices)
for iSample in range(da_nTrain): # radiation_noise，flip_augmentation
imdb_da_train['data'][:, :, :, iSample] = radiation_noise(
data[Row[da_RandPerm[iSample]] - HalfWidth: Row[da_RandPerm[iSample]] + HalfWidth + 1,
Column[da_RandPerm[iSample]] - HalfWidth: Column[da_RandPerm[iSample]] + HalfWidth + 1, :])
imdb_da_train['Labels'][iSample] = G[Row[da_RandPerm[iSample]], Column[da_RandPerm[iSample]]].astype(np.int64)
imdb_da_train['Labels'] = imdb_da_train['Labels'] - 1 # 将1-16变成0-15
imdb_da_train['set'] = np.ones([da_nTrain]).astype(np.int64)
print('ok')
return train_loader, test_loader, imdb_da_train, G, RandPerm, Row, Column, nTrain, nBand # , imdb_ori_train
#
test_IP_data = './datasets/IP/indian_pines_corrected.mat'
test_IP_label = './datasets/IP/indian_pines_gt.mat'
Data_Band_Scaler, GroundTruth = load_data(test_IP_data, test_IP_label)
def get_target_dataset(Data_Band_Scaler, GroundTruth):
train_loader, test_loader, imdb_da_train, G, RandPerm, Row, Column, nTrain,nBand= get_train_test_loader(Data_Band_Scaler = Data_Band_Scaler, GroundTruth = GroundTruth, \
class_num=TEST_CLASS_NUM, shot_num_per_class=TEST_SHOT_NUM_PER_CLASS+TEST_QUERY_NUM_PER_CLASS) # 9类，每类5个有标记样本
train_datas, train_labels = train_loader.__iter__().next()
print('train labels:', train_labels)
print('size of train datas:', train_datas.shape)
# print('train datas', train_datas[0, 0, :,:])
print(imdb_da_train.keys())
print(imdb_da_train['data'].shape) #(9, 9, 100, 225)
print(imdb_da_train['Labels'])
del Data_Band_Scaler#,GroundTruth
target_da_datas = np.transpose(imdb_da_train['data'], (3, 2, 0, 1)) # (9,9,100, 1800)->(1800, 100, 9, 9)
print(target_da_datas.shape)
target_da_labels = imdb_da_train['Labels'] # (1800,)
print('target data augmentation label:', target_da_labels)
target_da_train_set = {} # key:类别 value：列表 所有图片
for class_, path in zip(target_da_labels, target_da_datas): # labels：ndarray （5200，），data :ndarray (5200,3,64,64)
if class_ not in target_da_train_set:
target_da_train_set[class_] = []
target_da_train_set[class_].append(path) # path:ndarray (64,64,3)
target_da_metatrain_data = target_da_train_set # dice key：value 0：[]
print(target_da_metatrain_data.keys())
# print(metatrain_data[0][0].shape)
return train_loader, test_loader, target_da_metatrain_data, G, RandPerm, Row, Column, nTrain,nBand
class Task(object):
def __init__(self, data, num_classes, shot_num, query_num):
self.data = data
self.num_classes = num_classes
self.support_num = shot_num
self.query_num = query_num
class_folders = sorted(list(data))
class_list = random.sample(class_folders, self.num_classes)
labels = np.array(range(len(class_list)))
labels = dict(zip(class_list, labels))
samples = dict()
self.support_datas = []
self.query_datas = []
self.support_labels = []
self.query_labels = []
for c in class_list:
temp = self.data[c] # list
samples[c] = random.sample(temp, len(temp))
random.shuffle(samples[c])
self.support_datas += samples[c][:shot_num]
self.query_datas += samples[c][shot_num:shot_num + query_num]
self.support_labels += [labels[c] for i in range(shot_num)]
self.query_labels += [labels[c] for i in range(query_num)]
# print(self.support_labels)
# print(self.query_labels)
class FewShotDataset(Dataset):
def __init__(self, task, split='train'):
self.task = task
self.split = split
self.image_datas = self.task.support_datas if self.split == 'train' else self.task.query_datas
self.labels = self.task.support_labels if self.split == 'train' else self.task.query_labels
def __len__(self):
return len(self.image_datas)
def __getitem__(self, idx):
raise NotImplementedError("This is an abstract class. Subclass this class for your particular dataset.")
class HBKC_dataset(FewShotDataset):
def __init__(self, *args, **kwargs):
super(HBKC_dataset, self).__init__(*args, **kwargs)
def __getitem__(self, idx):
image = self.image_datas[idx]
label = self.labels[idx]
return image, label
# 采样器
class ClassBalancedSampler(Sampler):
''' Samples 'num_inst' examples each from 'num_cl' pool of examples of size 'num_per_class' '''
# 参数：
# num_per_class: 每个类的样本数量
# num_cl: 类别数量
# num_inst：support set或query set中的样本数量
# shuffle：样本是否乱序
def __init__(self, num_per_class, num_cl, num_inst,shuffle=True):
self.num_per_class = num_per_class
self.num_cl = num_cl
self.num_inst = num_inst
self.shuffle = shuffle
def __iter__(self):
# return a single list of indices, assuming that items will be grouped by class
if self.shuffle:
batch = [[i+j*self.num_inst for i in torch.randperm(self.num_inst)[:self.num_per_class]] for j in range(self.num_cl)]
else:
batch = [[i+j*self.num_inst for i in range(self.num_inst)[:self.num_per_class]] for j in range(self.num_cl)]
batch = [item for sublist in batch for item in sublist]
if self.shuffle:
random.shuffle(batch)
return iter(batch)
def __len__(self):
return 1
# dataloader
def get_HBKC_data_loader(task, num_per_class=1, split='train',shuffle = False):
# 参数:
# task: 当前任务
# num_per_class:每个类别的样本数量，与split有关
# split：‘train'或‘test'代表support和querya
# shuffle：样本是否乱序
# 输出：
# loader
dataset = HBKC_dataset(task,split=split)
if split == 'train':
sampler = ClassBalancedSampler(num_per_class, task.num_classes, task.support_num, shuffle=shuffle) # support set
else:
sampler = ClassBalancedSampler(num_per_class, task.num_classes, task.query_num, shuffle=shuffle) # query set
loader = DataLoader(dataset, batch_size=num_per_class*task.num_classes, sampler=sampler)
return loader
class FewShotDataset(Dataset):
def __init__(self, task, split='train'):
self.task = task
self.split = split
self.image_datas = self.task.support_datas if self.split == 'train' else self.task.query_datas
self.labels = self.task.support_labels if self.split == 'train' else self.task.query_labels
def __len__(self):
return len(self.image_datas)
def __getitem__(self, idx):
raise NotImplementedError("This is an abstract class. Subclass this class for your particular dataset.")
class HBKC_dataset(FewShotDataset):
def __init__(self, *args, **kwargs):
super(HBKC_dataset, self).__init__(*args, **kwargs)
def __getitem__(self, idx):
image = self.image_datas[idx]
label = self.labels[idx]
return image, label
# 采样器
class ClassBalancedSampler(Sampler):
''' Samples 'num_inst' examples each from 'num_cl' pool of examples of size 'num_per_class' '''
# 参数：
# num_per_class: 每个类的样本数量
# num_cl: 类别数量
# num_inst：support set或query set中的样本数量
# shuffle：样本是否乱序
def __init__(self, num_per_class, num_cl, num_inst,shuffle=True):
self.num_per_class = num_per_class
self.num_cl = num_cl
self.num_inst = num_inst
self.shuffle = shuffle
def __iter__(self):
# return a single list of indices, assuming that items will be grouped by class
if self.shuffle:
batch = [[i+j*self.num_inst for i in torch.randperm(self.num_inst)[:self.num_per_class]] for j in range(self.num_cl)]
else:
batch = [[i+j*self.num_inst for i in range(self.num_inst)[:self.num_per_class]] for j in range(self.num_cl)]
batch = [item for sublist in batch for item in sublist]
if self.shuffle:
random.shuffle(batch)
return iter(batch)
def __len__(self):
return 1
# dataloader
def get_HBKC_data_loader(task, num_per_class=1, split='train',shuffle = False):
# 参数:
# task: 当前任务
# num_per_class:每个类别的样本数量，与split有关
# split：‘train'或‘test'代表support和querya
# shuffle：样本是否乱序
# 输出：
# loader
dataset = HBKC_dataset(task,split=split)
if split == 'train':
sampler = ClassBalancedSampler(num_per_class, task.num_classes, task.support_num, shuffle=shuffle) # support set
else:
sampler = ClassBalancedSampler(num_per_class, task.num_classes, task.query_num, shuffle=shuffle) # query set
loader = DataLoader(dataset, batch_size=num_per_class*task.num_classes, sampler=sampler)
return loader
# MODEL
class VGG16_SE(nn.Module):
def __init__(self):
super(VGG16_SE, self).__init__()
self.preconv=nn.Conv2d(TAR_INPUT_DIMENSION,3,1,1,bias=False) # in_channels,out_channels,kernel_size,stride
self.preconv_bn=nn.BatchNorm2d(3)
self.conv1=nn.Sequential(
nn.Conv2d(3, 64, 3, padding=1),
nn.BatchNorm2d(64,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.conv2=nn.Sequential(
nn.Conv2d(64, 64, 3, padding=1),
nn.BatchNorm2d(64,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.maxpool2=nn.MaxPool2d(2, stride=2, ceil_mode=False)
self.conv3=nn.Sequential(
nn.Conv2d(64, 128, 3, padding=1),
nn.BatchNorm2d(128,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.conv4=nn.Sequential(
nn.Conv2d(128, 128, 3, padding=1),
nn.BatchNorm2d(128,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.maxpool4=nn.MaxPool2d(2, stride=2, ceil_mode=False)
self.conv5=nn.Sequential(
nn.Conv2d(128, 256, 3, padding=1),
nn.BatchNorm2d(256,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.conv6=nn.Sequential(
nn.Conv2d(256, 256, 3, padding=1),
nn.BatchNorm2d(256,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.conv7=nn.Sequential(
nn.Conv2d(256, 256, 3, padding=1),
nn.BatchNorm2d(256,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True),
)
self.maxpool7=nn.MaxPool2d(2, stride=2, ceil_mode=False)
self.conv=nn.Sequential(
nn.Conv2d(256,512,3,1),
nn.BatchNorm2d(512),
nn.ReLU(),
nn.MaxPool2d(2),
# nn.Dropout(0.5)
)
# self.fc=nn.Linear(512,TEST_CLASS_NUM)
self.fc=nn.Linear(512,100)#self.fc=nn.Linear(512,TEST_CLASS_NUM)
def forward(self, x): #torch.Size([9, 103, 33, 33])
x=self.preconv(x)
x=self.preconv_bn(x)
x = self.conv1(x)
x = self.conv2(x)
x = self.maxpool2(x)
x = self.conv3(x)
x = self.conv4(x)
x = self.maxpool4(x)
x = self.conv5(x)
x = self.conv6(x)
x = self.conv7(x)
x = self.maxpool7(x)
x=self.conv(x)
x= x.view(x.size(0), -1)
x=self.fc(x)
return x
class VGG16(nn.Module):
def __init__(self):
super(VGG16, self).__init__()
# self.preconv=nn.Conv2d(nBand,3,1,1,bias=False) # in_channels,out_channels,kernel_size,stride
# self.preconv_bn=nn.BatchNorm2d(3)
self.conv1=nn.Sequential(
nn.Conv2d(3, 64, 3, padding=1),
nn.BatchNorm2d(64,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.conv2=nn.Sequential(
nn.Conv2d(64, 64, 3, padding=1),
nn.BatchNorm2d(64,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.maxpool2=nn.MaxPool2d(2, stride=2, ceil_mode=False)
self.conv3=nn.Sequential(
nn.Conv2d(64, 128, 3, padding=1),
nn.BatchNorm2d(128,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.conv4=nn.Sequential(
nn.Conv2d(128, 128, 3, padding=1),
nn.BatchNorm2d(128,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.maxpool4=nn.MaxPool2d(2, stride=2, ceil_mode=False)
self.conv5=nn.Sequential(
nn.Conv2d(128, 256, 3, padding=1),
nn.BatchNorm2d(256,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.conv6=nn.Sequential(
nn.Conv2d(256, 256, 3, padding=1),
nn.BatchNorm2d(256,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True)
)
self.conv7=nn.Sequential(
nn.Conv2d(256, 256, 3, padding=1),
nn.BatchNorm2d(256,eps=1e-05,momentum=0.1, affine=True),
nn.ReLU(inplace=True),
)
self.maxpool7=nn.MaxPool2d(2, stride=2, ceil_mode=False)
self.conv=nn.Sequential(
nn.Conv2d(256,512,3,1),
nn.BatchNorm2d(512),
nn.ReLU(),
nn.MaxPool2d(2),
# nn.Dropout(0.5)
)
# self.fc = nn.Linear(512, TEST_CLASS_NUM)
self.fc = nn.Linear(512, 100)#self.fc=nn.Linear(512,TEST_CLASS_NUM)
def forward(self, x): #torch.Size([9, 103, 33, 33])
# x=self.preconv(x)
# x=self.preconv_bn(x)
x = self.conv1(x)
x = self.conv2(x)
x = self.maxpool2(x)
x = self.conv3(x)
x = self.conv4(x)
x = self.maxpool4(x)
x = self.conv5(x)
x = self.conv6(x)
x = self.conv7(x)
x = self.maxpool7(x)
x=self.conv(x)
x= x.view(x.size(0), -1)
x=self.fc(x)
return x
class AttentionLayer(nn.Module):
def __init__(self, channel, reduction=16):
super(AttentionLayer, self).__init__()
self.avg_pool = nn.AdaptiveAvgPool2d(1)
self.fc1=nn.Linear(channel, channel // reduction, bias=False)
self.relu=nn.ReLU(inplace=True)
self.fc2=nn.Linear(channel // reduction, channel, bias=False)
self.sigmoid=nn.Sigmoid()
def forward(self, x):
b, c, _, _ = x.size()
y = self.avg_pool(x).view(b, c)
y1=self.fc1(y)
self.fc1_feature=y1
y2=self.relu(y1)
self.relu_feature=y2
y3=self.fc2(y2)
self.fc2_feature=y3
y4=self.sigmoid(y3).view(b, c, 1, 1)
self.sigmoid_feature=y4
y5=x * y4.expand_as(x)
self.feature=y5
return y5
#1d cnn
class Spectral_1d(nn.Module):
def __init__(self,input_channels):
super(Spectral_1d, self).__init__()
self.feature_dim = input_channels
# Convolution Layer 1 kernel_size = (1, 1, 7), stride = (1, 1, 2), output channels = 24
self.conv1 = nn.Conv1d(1, 24, 7,stride=2,padding=0)
self.bn1 = nn.BatchNorm1d(24)
self.activation1 = nn.ReLU()
# Residual block 1
self.conv2 = nn.Conv1d(24,24,7,stride=1,padding=3)
self.bn2 = nn.BatchNorm1d(24)
self.activation2 = nn.ReLU()
self.conv3 = nn.Conv1d(24,24,7,stride=1,padding=3)
self.bn3 = nn.BatchNorm1d(24)
self.activation3 = nn.ReLU()
# Finish
# Convolution Layer 2 kernel_size = (1, 1, (self.feature_dim - 6) // 2), output channels = 128
self.conv4 = nn.Conv1d(24, 128, (self.feature_dim - 7) // 2 + 1,stride=1,padding=0)
self.bn5 = nn.BatchNorm1d(128)
self.activation5 = nn.ReLU()
self.fc1 = nn.Linear(in_features=128, out_features=100)
def forward(self,x):
# Convolution layer 1
x = self.conv1(x)
x = self.activation1(self.bn1(x))
# Residual layer 1
residual = x
x = self.conv2(x)
x = self.activation2(self.bn2(x))
x = self.conv3(x)
x = residual + x
x = self.activation3(self.bn3(x))
# Convolution layer 2
x = self.conv4(x)
x = self.activation5(self.bn5(x))
x = x.view(x.size()[0], -1)
x = self.fc1(x)
return x
def convert_vgg(vgg16):
net = VGG16()
vgg_items = net.state_dict().items()
vgg16_items = vgg16.items()
pretrain_model = {} # 字典 python
j = 0
i = 0
for k,v in list(net.state_dict().items()): # k='preconv.weight';v=torch.Size([3, 200, 1, 1])
v=list(vgg16_items)[j][1] #torch.Size([64, 3, 3, 3])
k=list(vgg_items)[i][0] #k=list(vgg_items)[i+6][0] #'preconv_bn.num_batches_tracked' 6 7 / 8 9 10 11 12
pretrain_model[k]=v
j = j+1
if j!=0 and j % 6 == 0:
i = i + 2
else:
i = i + 1
if j>=42:
break
return pretrain_model
def convert_vgg_se(vgg16):
net = VGG16_SE()
vgg_items = net.state_dict().items()
vgg16_items = vgg16.items()
# for key in vgg16:
# print(key, vgg16[key].size())
pretrain_model = {} # 字典 python
j = 0
# i = 0
for k,v in list(net.state_dict().items()): # k='preconv.weight';v=torch.Size([3, 200, 1, 1])
v=list(vgg16_items)[j][1] #torch.Size([64, 3, 3, 3])
k=list(vgg_items)[j+6][0] #k=list(vgg_items)[i+6][0] #'preconv_bn.num_batches_tracked' 6 7 / 8 9 10 11 12
pretrain_model[k]=v
j = j+1
# if j!=0 and j % 6 == 0:
# i = i + 2
# else:
# i = i + 1
if j>=49:
break
return pretrain_model
def euclidean_metric(a, b):
n = a.shape[0]
m = b.shape[0]
a = a.unsqueeze(1).expand(n, m, -1)
b = b.unsqueeze(0).expand(n, m, -1)
logits = -((a - b)**2).sum(dim=2)
return logits
############################miniImageNet###################################
# train miniImaget
print("Training miniageNet...")
# Step 1: init data folders
print("init data folders")
# init character folders for dataset construction
metatrain_folders, metatest_folders = tg.mini_imagenet_folders()
# model
model=VGG16()
print ('load the weight from vgg')# 加载模型
pretrained_dict = torch.load('./vgg16_bn-6c64b313.pth')
pretrained_dict = convert_vgg(pretrained_dict)
#自定义模型的 state_dict
model_dict = model.state_dict()
# 将model_pretrained的建与自定义模型的建进行比较，剔除不同的
pretrained_dict = {k: v for k, v in pretrained_dict.items() if k in model_dict}
# 更新现有的model_dict
model_dict.update(pretrained_dict)
# 加载我们真正需要的state_dict
model.load_state_dict(model_dict)
print ('copy the weight sucessfully')
# Loss and Optimizer
crossEntropy = nn.CrossEntropyLoss()
lr=0.001
optimizer = torch.optim.Adam(model.parameters(), lr=lr)
model.cuda()
last_accuracy = 0.0
best_episdoe = 0
train_loss = []
test_acc = []
running_D_loss, running_F_loss = 0.0, 0.0
running_label_loss = 0
running_domain_loss = 0
total_hit, total_num = 0.0, 0.0
test_acc_list = []
train_start = time.time()
for episode in range(1000): # EPISODE = 500000
# 取fsl 样本
task = tg.MiniImagenetTask(metatrain_folders, CLASS_NUM, SHOT_NUM_PER_CLASS, QUERY_NUM_PER_CLASS) # 5， 1，15
support_dataloader = tg.get_mini_imagenet_data_loader(task, num_per_class=SHOT_NUM_PER_CLASS, split="train", shuffle=False)
query_dataloader = tg.get_mini_imagenet_data_loader(task, num_per_class=QUERY_NUM_PER_CLASS, split="test", shuffle=True)
# sample datas
supports, support_labels = support_dataloader.__iter__().next() # (5, 100, 9, 9)
querys, query_labels = query_dataloader.__iter__().next() # (75,100,9,9)
# 计算特征,先计算特征，在混合
support_features = model(supports.cuda()) # torch.Size([409, 32, 7, 3, 3])
query_features = model(querys.cuda()) # torch.Size([409, 32, 7, 3, 3])
# 原型网络
if SHOT_NUM_PER_CLASS > 1:
support_proto = support_features.reshape(CLASS_NUM, SHOT_NUM_PER_CLASS, -1).mean(dim=1) # (9, 160)
else:
support_proto = support_features
# 分类损失
logits = euclidean_metric(query_features, support_proto)
loss = crossEntropy(logits, query_labels.cuda())
# 训练参数更新
model.zero_grad()
loss.backward()
optimizer.step()
total_hit += torch.sum(torch.argmax(logits, dim=1).cpu() == query_labels).item()
total_num += querys.shape[0]
if (episode + 1) % 100 == 0: # 没100次输出显示1次
# print('episode {:>3d}: transfer_loss: {:6.4f}, fsl_loss {:6.4f}, total_Loss {:6.4f},acc {:6.4f},'.format(
# episode + 1, transfer_loss.item(), f_loss.item(), loss.item(), total_hit / total_num))
print('episode {:>3d}: fsl_loss {:6.4f}, acc {:6.4f},'.format(
episode + 1, loss.item(), total_hit / total_num))
torch.save(model.state_dict(), "VGG16-IP.pth") # 只保存模型的参数
del metatrain_folders, metatest_folders
########################HSI################################
# model
class Network(nn.Module):
def __init__(self, n_bands, n_classes):
super(Network, self).__init__()
self.n_bands = n_bands
self.n_classes = n_classes
self.spat_model=VGG16_SE()
print ('load the weight from vgg')# 加载模型
# pretrained_dict = torch.load('/home/dell/lm/Transferdemo/Transfer demo/vgg16_bn-6c64b313.pth')
pretrained_dict = torch.load('./VGG16-IP.pth')
pretrained_dict = convert_vgg_se(pretrained_dict)
#自定义模型的 state_dict
model_dict = self.spat_model.state_dict()
# 将model_pretrained的建与自定义模型的建进行比较，剔除不同的
pretrained_dict = {k: v for k, v in pretrained_dict.items() if k in model_dict}
# 更新现有的model_dict
model_dict.update(pretrained_dict)
# 加载我们真正需要的state_dict
self.spat_model.load_state_dict(model_dict)
print ('copy the weight sucessfully')
# self.spat_model.conv1.add_module("AttentionLayer", AttentionLayer(64))
# self.spat_model.conv2.add_module("AttentionLayer", AttentionLayer(64))
# self.spat_model.conv3.add_module("AttentionLayer", AttentionLayer(128))
# self.spat_model.conv4.add_module("AttentionLayer", AttentionLayer(128))
# self.spat_model.conv5.add_module("AttentionLayer", AttentionLayer(256))
# self.spat_model.conv6.add_module("AttentionLayer", AttentionLayer(256))
# self.spat_model.conv7.add_module("AttentionLayer", AttentionLayer(256))
print (self.spat_model)
#Spectral_1d
self.spec_model = Spectral_1d(self.n_bands)
self.fc = nn.Linear(in_features=100 * 2, out_features=64, bias=False)
self.dropout = nn.Dropout(0.5)
self.classifier = nn.Linear(in_features=64, out_features=self.n_classes, bias=False)
# self.classifier = nn.Linear(in_features=100 * 2, out_features=self.n_classes, bias=False)
# self.classifier = nn.Linear(in_features=self.n_classes * 2, out_features=self.n_classes, bias=False)
def forward(self, x):
index = x.shape[2] // 2 # 4
train_data_spec = x[:, :, index, index] # (45, 100)
train_data_spec = train_data_spec.unsqueeze(dim=1) # ([64, 200,1, 17, 17])
spec_feature = self.spec_model(train_data_spec) ## (45, 32)
spat_feature = self.spat_model(x) ## (45, 64)
feature = torch.cat([spat_feature,spec_feature], 1)
feature = torch.relu(self.fc(feature))
feature = self.dropout(feature)
output = self.classifier(feature)
return output
nDataSet=2
acc = np.zeros([nDataSet, 1])
A = np.zeros([nDataSet, TEST_CLASS_NUM])
k = np.zeros([nDataSet, 1])
est_predict_all = []
best_acc_all = 0.0
best_G,best_RandPerm,best_Row, best_Column,best_nTrain = None,None,None,None,None
seeds = [1330, 1335, 1336, 1337, 1224, 1236, 1226, 1227, 1228, 1229] #
# seeds = [1333, 1335, 1336, 1337, 1224, 1236, 1226, 1235, 1233, 1229]
seeds = [1336,1224]
def train():
# np.random.seed(seeds[iDataSet])
train_loader, test_loader, target_da_metatrain_data, G, RandPerm, Row, Column, nTrain,nBand = get_target_dataset(
Data_Band_Scaler=Data_Band_Scaler, GroundTruth=GroundTruth)
model = Network(TAR_INPUT_DIMENSION, CLASS_NUM)
# Loss and Optimizer
crossEntropy = nn.CrossEntropyLoss()
lr = 0.001
optimizer = torch.optim.Adam(model.parameters(), lr=lr)
model.cuda()
print("HSI Training...")
last_accuracy = 0.0
best_episdoe = 0
train_loss = []
test_acc = []
running_D_loss, running_F_loss = 0.0, 0.0
running_label_loss = 0
running_domain_loss = 0
total_hit, total_num = 0.0, 0.0
test_acc_list = []