-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdiscriminator.py
executable file
·85 lines (58 loc) · 3.76 KB
/
discriminator.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
import tensorflow as tf
from tensorflow.contrib import slim
import numpy as np
def Discriminator(Target,DarkInput,reuse=None,is_training=True):
with tf.variable_scope('Discriminator',reuse=reuse):
with slim.arg_scope([slim.conv2d],padding='SAME',weights_regularizer=slim.l2_regularizer(0.001)):
def TargetConv(inputs=Target,reuse=reuse):
"""
The objective of this method is to simply take the Expected or
Gemerated target image and apply convolutions operations on it
adn bring it from a res of (None,16a,16b,3) to (None,a,b,1024)
"""
conv1 = slim.conv2d(inputs, 16, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv1 = slim.conv2d(conv1, 16, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv1 = slim.batch_norm(conv1, is_training=is_training)
conv2 = slim.conv2d(conv1, 32, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv2 = slim.conv2d(conv2, 32, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv2 = slim.batch_norm(conv2, is_training=is_training)
conv3 = slim.conv2d(conv2, 64, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv3 = slim.conv2d(conv3, 64, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv3 = slim.batch_norm(conv3, is_training=is_training)
conv4 = slim.conv2d(conv3, 128, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv4 = slim.conv2d(conv4, 128, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv4 = slim.batch_norm(conv4, is_training=is_training)
return conv4
def DarkInputConv(inputs=DarkInput,reuse=reuse):
"""
The objective of this method is to simply take Dark image
and apply convolutions operations on it
adn bring it from a res of (None,8a,8b,4) to (None,a,b,1024)
"""
conv2 = slim.conv2d(inputs, 32, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv2 = slim.conv2d(conv2, 32, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv2 = slim.batch_norm(conv2, is_training=is_training)
conv3 = slim.conv2d(conv2, 64, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv3 = slim.conv2d(conv3, 64, [3,3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv3 = slim.batch_norm(conv3, is_training=is_training)
conv4 = slim.conv2d(conv3, 128, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv4 = slim.conv2d(conv4, 128, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv4 = slim.batch_norm(conv4, is_training=is_training)
return conv4
Target = TargetConv(inputs=Target,reuse=reuse)
DarkInput = DarkInputConv(inputs=DarkInput,reuse=reuse)
combined = tf.concat([Target,DarkInput],axis=3)
conv5 = slim.conv2d(combined, 256, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv5 = slim.conv2d(conv5, 256, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv5 = slim.batch_norm(conv5, is_training=is_training)
conv6 = slim.conv2d(conv5, 256, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv6 = slim.conv2d(conv6, 256, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv6 = slim.batch_norm(conv6, is_training=is_training)
conv7 = slim.conv2d(conv6, 512, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=1)
conv7 = slim.conv2d(conv7, 512, [3, 3], rate=1, activation_fn=tf.nn.leaky_relu, stride=2)
#conv7 = slim.batch_norm(conv7, is_training=is_training)
DenseLayer=tf.reduce_mean(conv7,axis=[1,2])
DenseLayer=tf.layers.dense(inputs=DenseLayer,units=100,activation=tf.nn.leaky_relu)
DenseLayer = tf.layers.batch_normalization(DenseLayer)
DenseLayer=tf.layers.dense(inputs=DenseLayer,units=1,activation=None)
return DenseLayer,tf.nn.sigmoid(DenseLayer)