-
Notifications
You must be signed in to change notification settings - Fork 18
/
Copy pathinception_resnet_v1.py
186 lines (158 loc) · 8.76 KB
/
inception_resnet_v1.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
from models.imgclfmodel import ImgClfModel
from dataset.dataset import Dataset
import tensorflow as tf
from tensorflow.contrib.layers import conv2d
from tensorflow.contrib.layers import max_pool2d
from tensorflow.contrib.layers import avg_pool2d
from tensorflow.contrib.layers import flatten
from tensorflow.contrib.layers import fully_connected
class Inception_ResnetV1(ImgClfModel):
def __init__(self):
ImgClfModel.__init__(self, scale_to_imagenet=True)
def create_model(self, input):
with tf.variable_scope('stem'):
stem = conv2d(input, num_outputs=32,
kernel_size=[3,3], stride=2, padding='VALID',
activation_fn=tf.nn.relu)
stem = conv2d(stem, num_outputs=32,
kernel_size=[3,3], stride=1, padding='VALID',
activation_fn=tf.nn.relu)
stem = conv2d(stem, num_outputs=64,
kernel_size=[3,3], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
stem = max_pool2d(stem, kernel_size=[3,3], stride=2, padding='VALID')
stem = conv2d(stem, num_outputs=80,
kernel_size=[1,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
stem = conv2d(stem, num_outputs=192,
kernel_size=[3,3], stride=2, padding='VALID',
activation_fn=tf.nn.relu)
stem = conv2d(stem, num_outputs=256,
kernel_size=[3,3], stride=2, padding='VALID',
activation_fn=tf.nn.relu)
prev = stem
with tf.variable_scope('inception_resnet_a'):
for i in range(5):
identity = prev
branch_base = conv2d(prev, num_outputs=32,
kernel_size=[1,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_a = tf.layers.batch_normalization(branch_base)
branch_b = conv2d(branch_base, num_outputs=32,
kernel_size=[3,3], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_b = tf.layers.batch_normalization(branch_b)
branch_c = conv2d(branch_base, num_outputs=32,
kernel_size=[3,3], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_c = conv2d(branch_c, num_outputs=32,
kernel_size=[3,3], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_c = tf.layers.batch_normalization(branch_c)
layers_concat = list()
layers_concat.append(branch_a)
layers_concat.append(branch_b)
layers_concat.append(branch_c)
merge = tf.concat(layers_concat, 3)
merge = conv2d(merge, num_outputs=256,
kernel_size=[1,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
merge = tf.layers.batch_normalization(merge)
prev = tf.nn.relu(merge + identity)
prev = tf.layers.batch_normalization(prev)
with tf.variable_scope('reduction_a'):
branch_a = max_pool2d(prev, kernel_size=[3,3], stride=2, padding='VALID')
branch_b = conv2d(prev, num_outputs=384,
kernel_size=[3,3], stride=2, padding='VALID',
activation_fn=tf.nn.relu)
branch_c = conv2d(prev, num_outputs=192,
kernel_size=[1,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_c = conv2d(branch_c, num_outputs=192,
kernel_size=[3,3], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_c = conv2d(branch_c, num_outputs=256,
kernel_size=[3,3], stride=2, padding='VALID',
activation_fn=tf.nn.relu)
layers_concat = list()
layers_concat.append(branch_a)
layers_concat.append(branch_b)
layers_concat.append(branch_c)
prev = tf.concat(layers_concat, 3)
with tf.variable_scope('inception_resnet_b'):
for i in range(10):
identity = prev
branch_base = conv2d(prev, num_outputs=128,
kernel_size=[1,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_a = tf.layers.batch_normalization(branch_base)
branch_b = conv2d(branch_base, num_outputs=128,
kernel_size=[1,7], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_b = conv2d(branch_b, num_outputs=128,
kernel_size=[7,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_b = tf.layers.batch_normalization(branch_b)
layers_concat = list()
layers_concat.append(branch_a)
layers_concat.append(branch_b)
merge = tf.concat(layers_concat, 3)
merge = conv2d(merge, num_outputs=896,
kernel_size=[1,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
merge = tf.layers.batch_normalization(merge)
prev = tf.nn.relu(merge + identity)
prev = tf.layers.batch_normalization(prev)
with tf.variable_scope('reduction_b'):
branch_a = max_pool2d(prev, kernel_size=[3,3], stride=2, padding='VALID')
branch_base = conv2d(prev, num_outputs=256,
kernel_size=[1,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_b = conv2d(branch_base, num_outputs=384,
kernel_size=[3,3], stride=2, padding='VALID',
activation_fn=tf.nn.relu)
branch_c = conv2d(branch_base, num_outputs=256,
kernel_size=[3,3], stride=2, padding='VALID',
activation_fn=tf.nn.relu)
branch_d = conv2d(branch_base, num_outputs=256,
kernel_size=[3,3], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_d = conv2d(branch_d, num_outputs=256,
kernel_size=[3,3], stride=2, padding='VALID',
activation_fn=tf.nn.relu)
layers_concat = list()
layers_concat.append(branch_a)
layers_concat.append(branch_b)
layers_concat.append(branch_c)
layers_concat.append(branch_d)
prev = tf.concat(layers_concat, 3)
with tf.variable_scope('inception_resnet_c'):
for i in range(5):
identity = prev
branch_base = conv2d(prev, num_outputs=192,
kernel_size=[1,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_a = tf.layers.batch_normalization(branch_base)
branch_b = conv2d(branch_base, num_outputs=192,
kernel_size=[1,3], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_b = conv2d(branch_b, num_outputs=192,
kernel_size=[3,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
branch_b = tf.layers.batch_normalization(branch_b)
layers_concat = list()
layers_concat.append(branch_a)
layers_concat.append(branch_b)
merge = tf.concat(layers_concat, 3)
merge = conv2d(merge, num_outputs=1792,
kernel_size=[1,1], stride=1, padding='SAME',
activation_fn=tf.nn.relu)
merge = tf.layers.batch_normalization(merge)
prev = tf.nn.relu(merge + identity)
prev = tf.layers.batch_normalization(prev)
with tf.variable_scope('final'):
prev = avg_pool2d(prev, kernel_size=[3,3], stride=2, padding='SAME')
flat = flatten(prev)
dr = tf.nn.dropout(flat, 0.8)
self.out = fully_connected(dr, num_outputs=self.num_classes, activation_fn=None)
return [self.out]