-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathMain.py
130 lines (122 loc) · 6.21 KB
/
Main.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
# 1. 数据预处理，包括读数据，生成数组文件等等
#
import argparse
from Preprocess import Preprocess,PerClassSplit,splitTrainTestSet
from TrainAE import TrainAAE_patch,SaveFeatures_AAE,TrainVAE_patch,SaveFeatures_VAE
from TrainPCL import TrainContrast, ContrastPredict
import numpy as np
import joblib
from TrainSVM import TrainSVM,TestSVM, TrainNN, TestNN
from sklearn.metrics import confusion_matrix, accuracy_score, classification_report, cohen_kappa_score
import time
import csv
from utils import reports
dataset_names = ['IP','SA','PU']
parser = argparse.ArgumentParser(description="Run deep learning experiments on"
" various hyperspectral datasets")
parser.add_argument('--dataset', type=str, default='SA', choices=dataset_names,
help="Dataset to use.")
parser.add_argument('--train',type=int, default=0,choices=(0,1))
parser.add_argument('--perclass', type=float, default=5) #会除以100
parser.add_argument('--device', type=str, default="cuda:0", choices=("cuda:0","cuda:1"))
parser.add_argument('--projection_dim', type=int, default=128)
parser.add_argument('--encoded_dim', type=int, default=128)
parser.add_argument('--batch_size', type=int, default=128)
parser.add_argument('--temperature', type=float, default=1)#程序中会除以100
parser.add_argument('--Windowsize', type=int, default=27)
parser.add_argument('--classifier', type=str, default='linear',choices=("linear","svm"))
parser.add_argument('--train_contrast',type=int, default=0,choices=(0,1))
parser.add_argument('--Patch_channel', type=int, default=15)
parser.add_argument('--RandomSeed', type=bool, default=False)#True 代表固定随机数， False代表不固定
args = parser.parse_args()
if args.RandomSeed:
randomState=345
else:
randomState=int(np.random.randint(1, high=1000))
args.temperature=args.temperature/100
args.perclass=args.perclass/100
print(args)
output_units = 9 if (args.dataset == 'PU' or args.dataset == 'PC') else 16
Datadir='./DataArray/'
XPath = Datadir + 'X.npy'
yPath = Datadir + 'y.npy'
# # 2. 生成两个自编码器并训练，保存好模型参数
train_start = time.time()
if args.train:
Preprocess(XPath, yPath, args.dataset, args.Windowsize, Patch_channel=args.Patch_channel)
TrainVAE_patch(XPath, Patch_channel=args.Patch_channel, windowSize=args.Windowsize, encoded_dim=args.encoded_dim,
batch_size=args.batch_size)
TrainAAE_patch(XPath,Patch_channel=args.Patch_channel,windowSize=args.Windowsize,encoded_dim=args.encoded_dim,batch_size=args.batch_size)
# # 3. 用已保存的自编码器将两种数据编码成两个特征数组存起来
test_start1 = time.time()
VAEPath=Datadir+'VAE_Features.npy'
AAEPath=Datadir+'AAE_Features.npy'
VAEFeatures=SaveFeatures_VAE(XPath,Patch_channel=args.Patch_channel,windowSize=args.Windowsize,encoded_dim=args.encoded_dim,batch_size=args.batch_size)
np.save(VAEPath,VAEFeatures)
AAEFeatures=SaveFeatures_AAE(XPath,Patch_channel=args.Patch_channel,windowSize=args.Windowsize,encoded_dim=args.encoded_dim,batch_size=args.batch_size)
np.save(AAEPath,AAEFeatures)
test_stop1 = time.time()
#############训练对比网络
if args.train_contrast==True:
TrainContrast(AAEPath,VAEPath,batch_size=128,temperature=args.temperature,projection_dim=args.projection_dim,input_dim=args.encoded_dim)
train_stop = time.time()
test_start2 = time.time()
ContrastFeature=ContrastPredict(AAEPath,VAEPath,batch_size=128,temperature=args.temperature,projection_dim=args.projection_dim,input_dim=args.encoded_dim)
test_stop2 = time.time()
np.save(Datadir+'ContrastFeature.npy',ContrastFeature)
y = np.load(yPath)
stratify = np.arange(0, output_units, 1)
for feature in ["Contrast","AAE","VAE"]:
if feature == "Contrast":
fea = ContrastFeature
projection_dim=args.projection_dim
elif feature == "AAE":
fea = AAEFeatures
projection_dim=1024
elif feature == "VAE":
fea = VAEFeatures
projection_dim = 1024
if args.perclass > 1:
Xtrain, Xtest, ytrain, ytest = PerClassSplit(fea, y, args.perclass, stratify,randomState=randomState)
else:
Xtrain, Xtest, ytrain, ytest = splitTrainTestSet(fea, y, 1 - args.perclass,randomState=randomState)
np.save(Datadir + 'Xtrain.npy', Xtrain)
np.save(Datadir + 'ytrain.npy', ytrain)
np.save(Datadir + 'Xtest.npy', Xtest)
np.save(Datadir + 'ytest.npy', ytest)
if args.classifier=='svm':
train_start2 = time.time()
SVM_model=TrainSVM(Xtrain,ytrain)
train_stop2 = time.time()
joblib.dump(SVM_model, './models/SVM.model')
SVM_model=joblib.load('./models/SVM.model')
test_start3 = time.time()
Predictions=TestSVM(Xtest,SVM_model)
test_stop3 = time.time()
else:
train_start2 = time.time()
ModelPath=TrainNN(n_features=projection_dim,n_classes=output_units,Datadir=Datadir)
train_stop2 = time.time()
test_start3 = time.time()
Predictions=TestNN(n_features=projection_dim,n_classes=output_units, ModelPath=ModelPath,Datadir=Datadir)
test_stop3 = time.time()
#####算上训练分类器和测试分类器的时间
# TrainTime=(train_stop-train_start)+(train_stop2-train_start2)
# TestTime=(test_stop1-test_start1)+(test_stop2-test_start2)+(test_stop3-test_start3)
######不计分类器的训练和测试时间损耗
TrainTime = (train_stop - train_start)
TestTime=(test_stop1-test_start1)+(test_stop2-test_start2)
ytest=np.load(Datadir+'ytest.npy')
classification = classification_report(ytest.astype(int), Predictions)
print(classification)
classification, confusion, oa, each_acc, aa, kappa = reports(Predictions, ytest.astype(int), args.dataset)
## 9. 存储分类报告至csv，以及画出效果图
# file_name = "AECLR_IP"+'_'+str(perclass)+"perclass.txt"
# with open(file_name, 'w') as x_file:
# x_file.write('{}'.format(classification))
each_acc_str = ','.join(str(x) for x in each_acc)
add_info=[args.dataset,args.perclass,args.temperature,args.Windowsize,args.classifier,feature,oa,aa,kappa,TrainTime,TestTime]+each_acc_str.split('[')[0].split(']')[0].split(',')
csvFile = open("compare.csv", "a")
writer = csv.writer(csvFile)
writer.writerow(add_info)
csvFile.close()