test_model.py
6.06 KB
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
__author__ = 'chunk'
from sklearn import cross_validation
from ..common import *
from ..mdata import CV, ILSVRC, ILSVRC_S
from ..mmodel.svm import SVM
from ..mmodel.theano import THEANO
import gzip
import cPickle
timer = Timer()
package_dir = os.path.dirname(os.path.abspath(__file__))
def test_SVM_CV():
timer.mark()
dcv = CV.DataCV()
X, Y = dcv.load_data(mode='local') # 90.468586s -> 5.392520s
# X, Y = dcv.load_data(mode='hbase') # 21.682754s
# X, Y = dcv.load_data(mode='spark') # 29.549597s
timer.report()
timer.mark()
# msvm = SVM.ModelSVM(toolset='sklearn') # 3.030380s
# msvm = SVM.ModelSVM(toolset='opencv') # 8.939880s
# msvm = SVM.ModelSVM(toolset='libsvm') # 185.524023s
msvm = SVM.ModelSVM(toolset='spark')
msvm.train(X, Y)
timer.report()
timer.mark()
for path, subdirs, files in os.walk('data/467/'):
for name in files:
imgpath = os.path.join(path, name)
feat = dcv.get_feat(imgpath, 'hog')
print name, msvm.predict(feat)
timer.report()
timer.mark()
print msvm.test(X, Y) # 0.948892561983 for svm_cv, 0.989024793388 for svm_sk, 0.9900826446280992 for svm_lib
timer.report() # 27.421949s for svm_lib
def test_SVM_ILSVRC():
timer.mark()
dil = ILSVRC.DataILSVRC(base_dir='/data/hadoop/ImageNet/ILSVRC/ILSVRC2013_DET_val', category='Train_5000_0.05_orig')
X, Y = dil.load_data(mode='local') #
# X, Y = dil.load_data(mode='hbase') #
# X, Y = dil.load_data(mode='spark') #
X_train, X_test, Y_train, Y_test = cross_validation.train_test_split(X, Y, test_size=0.4, random_state=0)
print np.array(Y).shape, np.array(X).shape
print np.array(X_train).shape, np.array(Y_train).shape
print np.array(X_test).shape, np.array(Y_test).shape
timer.report()
timer.mark()
msvm = SVM.ModelSVM(toolset='sklearn') # 4.884247s 0.777853030816
# msvm = SVM.ModelSVM(toolset='opencv') #
# msvm = SVM.ModelSVM(toolset='libsvm') #
# msvm = SVM.ModelSVM(toolset='spark')
msvm.train(X_train, Y_train)
timer.report()
timer.mark()
print msvm.test(X_test, Y_test) #
timer.report() #
# timer.mark()
# print 'or like this:'
# scores = cross_validation.cross_val_score(msvm.model, X, Y)
# print scores
# timer.report()
def test_SVM_ILSVRC_HBASE():
timer.mark()
# dil = ILSVRC.DataILSVRC(base_dir='ILSVRC2013_DET_val', category='Train_3')
# X, Y = dil.load_data(mode='hbase') # pass
dils = ILSVRC_S.DataILSVRC_S(base='ILSVRC2013_DET_val', category='Train_5000')
X, Y = dils.load_data(mode='hbase') # pass
dil = ILSVRC_S.DataILSVRC_S(base='/data/hadoop/ImageNet/ILSVRC/ILSVRC2013_DET_val/', category='Train_5000_0.1_orig')
X1, Y1 = dil.load_data(mode='local')
X_train, X_test, Y_train, Y_test = cross_validation.train_test_split(X, Y, test_size=0.4, random_state=0)
print Y, np.sum(np.array(Y) == 0), np.sum(np.array(Y) == 1)
print np.array(Y).shape, np.array(X).shape
print np.array(X_train).shape, np.array(Y_train).shape
print np.array(X_test).shape, np.array(Y_test).shape
timer.report()
timer.mark()
msvm = SVM.ModelSVM(toolset='sklearn') # 4.884247s 0.777853030816
# msvm = SVM.ModelSVM(toolset='opencv') #
# msvm = SVM.ModelSVM(toolset='libsvm') #
# msvm = SVM.ModelSVM(toolset='spark')
msvm.train(X_train, Y_train)
timer.report()
timer.mark()
print msvm.test(X_test, Y_test) #
timer.report() #
timer.mark()
print msvm.test(X1, Y1) #(0.048868415782094936, 0.4924709948160948, 0.74568774878372401)
timer.report() #
# timer.mark()
# print 'or like this:'
# scores = cross_validation.cross_val_score(msvm.model, X, Y)
# print scores
# timer.report()
def test_SVM_ILSVRC_TEST():
timer.mark()
dil = ILSVRC_S.DataILSVRC_S(base='/data/hadoop/ImageNet/ILSVRC/ILSVRC2013_DET_val/', category='Train_5000_0.1_orig')
X1, Y1 = dil.load_data(mode='local')
timer.report()
timer.mark()
msvm = SVM.ModelSVM(toolset='sklearn') # 4.884247s 0.777853030816
timer.report()
timer.mark()
print msvm.test(X1, Y1) #(0.048868415782094936, 0.4924709948160948, 0.74568774878372401)
timer.report() #
# timer.mark()
# print 'or like this:'
# scores = cross_validation.cross_val_score(msvm.model, X, Y)
# print scores
# timer.report()
def test_SVM_ILSVRC_SPARK():
timer.mark()
dils = ILSVRC_S.DataILSVRC_S(base='ILSVRC2013_DET_val', category='Test_1')
rdd_dataset = dils.load_data(mode='spark') # pass
timer.report()
timer.mark()
# msvm = SVM.ModelSVM(toolset='sklearn') #
# msvm = SVM.ModelSVM(toolset='opencv') #
# msvm = SVM.ModelSVM(toolset='libsvm') #
msvm = SVM.ModelSVM(toolset='spark', sc=dils.sparker)
msvm.train(rdd_dataset)
timer.report()
dataset = rdd_dataset.collect()
length = len(dataset)
X_test, Y_test = [dataset[i].features for i in range(length)], [dataset[i].label for i in range(length)]
timer.mark()
print msvm.test(dils.sparker.sc.parallelize(X_test), Y_test) #
timer.report() #
def test_SVM_ILSVRC_S():
test_SVM_ILSVRC_HBASE()
# test_SVM_ILSVRC_SPARK()
def test_THEANO_mnist():
mtheano = THEANO.ModelTHEANO(toolset='cnn')
mtheano._train_cnn(learning_rate=0.1, n_epochs=200, dataset=os.path.join(package_dir, '../res/', 'mnist.pkl.gz'), nkerns=[20, 50], batch_size=500)
def test_THEANO_crop():
timer.mark()
dilc = ILSVRC.DataILSVRC(base_dir='/data/hadoop/ImageNet/ILSVRC/ILSVRC2013_DET_val', category='Train_5000_crop_pil')
X, Y = dilc.load_data(mode='local', feattype='coef')
print X[0],Y
timer.report()
# X_train, X_test, Y_train, Y_test = cross_validation.train_test_split(X, Y, test_size=0.2, random_state=0)
# with open(os.path.join(package_dir,'../res/','ils_crop.pkl'),'wb') as f:
# cPickle.dump([(X_train,Y_train),(X_test,Y_test)], f)
timer.mark()
mtheano = THEANO.ModelTHEANO(toolset='cnn')
mtheano._train_cnn(X, Y)
timer.report()
if __name__ == '__main__':
# test_SVM_CV()
test_SVM_ILSVRC()
print 'helllo'