-
Notifications
You must be signed in to change notification settings - Fork 4
/
vgg16.py
135 lines (100 loc) · 5.16 KB
/
vgg16.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
from __future__ import division
from __future__ import print_function
from __future__ import absolute_import
import tensorflow as tf
import numpy as np
VGG_MEAN = [103.939, 116.779, 123.68]
class Vgg16(object):
"""docstring for Vgg16"""
def __init__(self,vgg16_npy_path=None,trainable=True):
super(Vgg16, self).__init__()
if vgg16_npy_path is not None:
self.data_dict = np.load(vgg16_npy_path, encoding='latin1').item()
else:
self.data_dict = None
self.trainable = trainable
def forward(self,images):
self.conv1_1 = self.conv_layer(images, 3, 64, "conv1_1")
self.conv1_2 = self.conv_layer(self.conv1_1, 64, 64, "conv1_2")
self.pool1 = self.max_pool(self.conv1_2, 'pool1')
self.conv2_1 = self.conv_layer(self.pool1, 64, 128, "conv2_1")
self.conv2_2 = self.conv_layer(self.conv2_1, 128, 128, "conv2_2")
self.pool2 = self.max_pool(self.conv2_2, 'pool2')
self.conv3_1 = self.conv_layer(self.pool2, 128, 256, "conv3_1")
self.conv3_2 = self.conv_layer(self.conv3_1, 256, 256, "conv3_2")
self.conv3_3 = self.conv_layer(self.conv3_2, 256, 256, "conv3_3")
self.pool3 = self.max_pool(self.conv3_3, 'pool3')
self.conv4_1 = self.conv_layer(self.pool3, 256, 512, "conv4_1")
self.conv4_2 = self.conv_layer(self.conv4_1, 512, 512, "conv4_2")
self.conv4_3 = self.conv_layer(self.conv4_2, 512, 512, "conv4_3")
self.pool4 = self.max_pool(self.conv4_3, 'pool4')
self.conv5_1 = self.conv_layer(self.pool4, 512, 512, "conv5_1")
self.conv5_2 = self.conv_layer(self.conv5_1, 512, 512, "conv5_2")
self.conv5_3 = self.conv_layer(self.conv5_2, 512, 512, "conv5_3")
self.pool5 = self.max_pool(self.conv5_3, 'pool5')
self.conv6_1 = self.det_conv_layer(self.pool5,512,256,"conv6_1")
self.conv7_1 = self.det_conv_layer(self.conv6_1,256,256,"conv7_1")
self.conv8_1 = self.det_conv_layer(self.conv7_1,256,256,"conv8_1")
self.conv_end = self.det_conv_layer(self.conv8_1,256,30,'conv9_1',relu=False)
sigmoid_out = tf.sigmoid(self.conv_end,name='output')
return sigmoid_out
def det_conv_layer(self,bottom,in_channels,out_channels,name,relu=True):
with tf.variable_scope(name):
residual = tf.identity(bottom)
filt, conv_biases = self.get_conv_var(3, in_channels, out_channels, name+"_1")
conv = tf.nn.conv2d(bottom, filt, [1, 1, 1, 1], padding='SAME')
bias = tf.nn.bias_add(conv, conv_biases)
bias = tf.layers.batch_normalization(bias,training=self.trainable,name=name+'_bn1')
filt2,conv_biases2 = self.get_conv_var(3,out_channels,out_channels,name+"_2")
conv2 = tf.nn.conv2d(bias,filt2,[1,1,1,1],padding='SAME')
bias2 = tf.nn.bias_add(conv2,conv_biases2)
bias2 = tf.layers.batch_normalization(bias2,training=self.trainable,name=name+'_bn2')
if in_channels!=out_channels:
filt3,conv_biases3 = self.get_conv_var(3,in_channels,out_channels,name+'_3')
conv3 = tf.nn.conv2d(residual,filt3,[1,1,1,1],padding='SAME')
bias3 = tf.nn.bias_add(conv3,conv_biases3)
residual = tf.layers.batch_normalization(bias3,training=self.trainable,name=name+'_bn3')
out = tf.nn.relu(residual+bias2) if relu else (residual+bias2)
return out
def max_pool(self, bottom, name):
return tf.nn.max_pool(bottom, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME', name=name)
def conv_layer(self, bottom, in_channels, out_channels, name):
with tf.variable_scope(name):
filt, conv_biases = self.get_conv_var(3, in_channels, out_channels, name)
conv = tf.nn.conv2d(bottom, filt, [1, 1, 1, 1], padding='SAME')
bias = tf.nn.bias_add(conv, conv_biases)
relu = tf.nn.relu(bias)
return relu
def get_conv_var(self, filter_size, in_channels, out_channels, name):
initial_value = tf.truncated_normal([filter_size, filter_size, in_channels, out_channels], 0.0, 0.001)
filters = self.get_var(initial_value, name, 0, name + "_filters")
initial_value = tf.truncated_normal([out_channels], .0, .001)
biases = self.get_var(initial_value, name, 1, name + "_biases")
weight_decay = tf.multiply(tf.nn.l2_loss(filters),5e-4,name='weight_loss')
tf.add_to_collection('losses',weight_decay)
bias_decay = tf.multiply(tf.nn.l2_loss(biases),5e-4,name='bias_loss')
tf.add_to_collection('losses',bias_decay)
return filters, biases
def get_var(self, initial_value, name, idx, var_name):
if self.data_dict is not None and name in self.data_dict:
value = self.data_dict[name][idx]
else:
value = initial_value
var = tf.Variable(value, name=var_name)
# print var_name, var.get_shape().as_list()
assert var.get_shape() == initial_value.get_shape()
return var
if __name__=='__main__':
vgg16 = Vgg16(vgg16_npy_path='./model/vgg16.npy')
with tf.device('/cpu:0'):
data = np.random.randn(64,448,448,3)
inputs = tf.placeholder(tf.float32,[64,448,448,3])
predicts = vgg16.forward(inputs)
config = tf.ConfigProto(allow_soft_placement=True)
config.gpu_options.allow_growth=True
init = tf.global_variables_initializer()
sess = tf.Session(config=config)
sess.run(init)
output = sess.run(predicts,feed_dict={inputs:data})
print(output.shape)
sess.close()