-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathface_train_resnet50.py
213 lines (168 loc) · 8.07 KB
/
face_train_resnet50.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
#-*- coding: utf-8 -*-
import random
from keras.layers import Input,merge,BatchNormalization
from keras.layers import Conv2D, MaxPooling2D, ZeroPadding2D, AveragePooling2D
from keras.models import Model
from sklearn.cross_validation import train_test_split
from keras.layers import Dense, Activation, Flatten
from keras.optimizers import SGD
from keras.utils import np_utils
from keras.models import load_model
from keras import backend as K
from load_face import load_train_orl,load_test_orl,load_train_pie,load_test_pie
model=None
class Dataset:
def __init__(self, path_name):
self.train_images = None
self.train_labels = None
self.valid_images = None
self.valid_labels = None
self.test_images = None
self.test_labels = None
self.path_name = path_name
self.input_shape = None
def load(self, img_rows = 64, img_cols = 64,
img_channels = 1, nb_classes = 68):
images, labels = load_train_pie(self.path_name)
train_images, valid_images, train_labels, valid_labels = train_test_split(images, labels, test_size = 0.2, random_state = random.randint(0, 100))
test_images,test_labels = load_test_pie(self.path_name)
#当前的维度顺序如果为'th'，则输入图片数据时的顺序为：channels,rows,cols，否则:rows,cols,channels
#这部分代码就是根据keras库要求的维度顺序重组训练数据集
if K.image_dim_ordering() == 'th':
train_images = train_images.reshape(train_images.shape[0], img_channels, img_rows, img_cols)
valid_images = valid_images.reshape(valid_images.shape[0], img_channels, img_rows, img_cols)
test_images = test_images.reshape(test_images.shape[0], img_channels, img_rows, img_cols)
self.input_shape = (img_channels, img_rows, img_cols)
else:
train_images = train_images.reshape(train_images.shape[0], img_rows, img_cols, img_channels)
valid_images = valid_images.reshape(valid_images.shape[0], img_rows, img_cols, img_channels)
test_images = test_images.reshape(test_images.shape[0], img_rows, img_cols, img_channels)
self.input_shape = (img_rows, img_cols, img_channels)
print(train_images.shape[0], 'train samples')
print(valid_images.shape[0], 'valid samples')
print(test_images.shape[0], 'test samples')
train_labels = np_utils.to_categorical(train_labels, nb_classes)
valid_labels = np_utils.to_categorical(valid_labels, nb_classes)
test_labels = np_utils.to_categorical(test_labels, nb_classes)
train_images = train_images.astype('float32')
valid_images = valid_images.astype('float32')
test_images = test_images.astype('float32')
train_images /= 255
valid_images /= 255
test_images /= 255
self.train_images = train_images
self.valid_images = valid_images
self.test_images = test_images
self.train_labels = train_labels
self.valid_labels = valid_labels
self.test_labels = test_labels
def identity_block(input_tensor, kernel_size, filters):
dim_ordering = K.image_dim_ordering()
nb_filter1, nb_filter2, nb_filter3 = filters
if dim_ordering == 'tf':
axis = 3
else:
axis = 1
out = Conv2D(nb_filter1, (1, 1), data_format="channels_last")(input_tensor)
out = BatchNormalization(axis=axis)(out)
out = Activation('relu')(out)
out = out = Conv2D(nb_filter2, (kernel_size, kernel_size), padding='same',
data_format="channels_last")(out)
out = BatchNormalization(axis=axis)(out)
out = Activation('relu')(out)
out = Conv2D(nb_filter3, (1, 1), data_format="channels_last")(out)
out = BatchNormalization(axis=axis)(out)
out = merge([out, input_tensor], mode='sum')
out = Activation('relu')(out)
return out
def conv_block(input_tensor, kernel_size, filters, strides=(2, 2)):
nb_filter1, nb_filter2, nb_filter3 = filters
dim_ordering = K.image_dim_ordering()
if dim_ordering == 'tf':
axis = 3
else:
axis = 1
out = Conv2D(nb_filter1, (1, 1), strides=strides,data_format="channels_last")(input_tensor)
out = BatchNormalization(axis=axis)(out)
out = Activation('relu')(out)
out = Conv2D(nb_filter2, (kernel_size, kernel_size), padding='same',data_format="channels_last")(out)
out = BatchNormalization(axis=axis)(out)
out = Activation('relu')(out)
out = Conv2D(nb_filter3, (1, 1), data_format="channels_last")(out)
out = BatchNormalization(axis=axis)(out)
shortcut = Conv2D(nb_filter3, (1, 1), strides=strides,data_format="channels_last")(input_tensor)
shortcut = BatchNormalization(axis=axis)(shortcut)
out = merge([out, shortcut], mode='sum')
out = Activation('relu')(out)
return out
def get_resnet50(dataset):
if K.image_dim_ordering() == 'tf':
axis = 3
else:
axis = 1
inp = Input(shape=dataset.input_shape)
dim_ordering = K.image_dim_ordering()
out = ZeroPadding2D((3, 3), data_format="channels_last")(inp)
out = Conv2D(64, (7, 7), strides=(2, 2), data_format="channels_last")(out)
out = BatchNormalization(axis=axis)(out)
out = Activation('relu')(out)
out = MaxPooling2D((3, 3), strides=(2, 2), data_format="channels_last")(out)
out = conv_block(out, 3, [64, 64, 256], strides=(1, 1))
out = identity_block(out, 3, [64, 64, 256])
out = identity_block(out, 3, [64, 64, 256])
out = conv_block(out, 3, [128, 128, 512])
out = identity_block(out, 3, [128, 128, 512])
out = identity_block(out, 3, [128, 128, 512])
out = identity_block(out, 3, [128, 128, 512])
out = conv_block(out, 3, [256, 256, 1024])
out = identity_block(out, 3, [256, 256, 1024])
out = identity_block(out, 3, [256, 256, 1024])
out = identity_block(out, 3, [256, 256, 1024])
out = identity_block(out, 3, [256, 256, 1024])
out = identity_block(out, 3, [256, 256, 1024])
out = conv_block(out, 3, [512, 512, 2048])
out = identity_block(out, 3, [512, 512, 2048])
out = identity_block(out, 3, [512, 512, 2048])
out = AveragePooling2D((2, 2), data_format="channels_last")(out)
out = Flatten()(out)
out = Dense(68, activation='softmax')(out)
model = Model(inp, out)
model.summary()
return model
def save_mod(file_path):
model.save(file_path)
def load_mod(file_path):
model = load_model(file_path)
def evaluate(dataset):
score = model.evaluate(dataset.test_images, dataset.test_labels, verbose = 1)
print("%s: %.2f%%" % (model.metrics_names[1], score[1] * 100))
#训练模型
def train(dataset, batch_size = 40, nb_epoch = 100):
sgd = SGD(lr = 0.01, decay = 1e-6,
momentum = 0.9, nesterov = True) #采用SGD+momentum的优化器进行训练，首先生成一个优化器对象
model.compile(loss='categorical_crossentropy',
optimizer=sgd,
metrics=['accuracy']) #完成实际的模型配置工作
model.fit(dataset.train_images,
dataset.train_labels,
batch_size = batch_size,
epochs = nb_epoch,
validation_data = (dataset.valid_images, dataset.valid_labels),
shuffle = True)
if __name__ == '__main__':
dataset = Dataset('data/PIE dataset/')
dataset.load()
#train
model = get_resnet50(dataset)
sgd = SGD(lr = 0.01, decay = 1e-6,
momentum = 0.9, nesterov = True)
model.compile(loss='categorical_crossentropy',
optimizer=sgd,
metrics=['accuracy'])
model.load_weights('face_model_resnet50.h5')
# train(dataset)
# save_mod('face_model_pie.h5')
#test
# load_mod(file_path = '123.h5')
print("loaded")
evaluate(dataset)