-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathunet.py
79 lines (63 loc) · 4.46 KB
/
unet.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
import tensorflow as tf
from tensorflow.keras import backend as K
from tensorflow.keras.layers import *
from tensorflow.keras.optimizers import Adam
def unet(input_size = (256, 256, 1)):
initializer = 'he_normal'
inputs = Input(shape=input_size) #(240,240,1)
conv1 = Conv2D(32, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(inputs) #(240,240,32)
conv1 = Conv2D(32, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(conv1) #(240,240,32)
pool1 = MaxPooling2D(pool_size=(2, 2))(conv1) #(120,120,32)
conv2 = Conv2D(64, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(pool1) #(120,120,64)
conv2 = Conv2D(64, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(conv2) #(120,120,64)
pool2 = MaxPooling2D(pool_size=(2, 2))(conv2) #(60,60,64)
conv3 = Conv2D(128, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(pool2) #(60,60,128)
conv3 = Conv2D(128, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(conv3) #(60,60,128)
pool3 = MaxPooling2D(pool_size=(2, 2))(conv3) #(30,30,128)
conv4 = Conv2D(256, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(pool3) #(30,30,256)
conv4 = Conv2D(256, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(conv4) #(30,30,256)
pool4 = MaxPooling2D(pool_size=(2, 2))(conv4) #(15,15,256)
conv5 = Conv2D(512, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(pool4) #(15,15,512)
conv5 = Conv2D(512, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(conv5) #(15,15,512)
up6 = concatenate([Conv2DTranspose(256, (2, 2), strides=(2, 2), padding='same', #(30,30,256)
kernel_initializer=initializer)(conv5),conv4], axis=3) #(30,30,512)
conv6 = Conv2D(256, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(up6) #(30,30,256)
conv6 = Conv2D(256, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(conv6) #(30,30,256)
up7 = concatenate([Conv2DTranspose(128, (2, 2), strides=(2, 2), padding='same', #(60,60,128)
kernel_initializer=initializer)(conv6),conv3], axis=3) #(60,60,256)
conv7 = Conv2D(128, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(up7) #(60,60,128)
conv7 = Conv2D(128, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(conv7) #(60,60,128)
up8 = concatenate([Conv2DTranspose(64, (2, 2), strides=(2, 2),padding='same', #(120,120,64)
kernel_initializer=initializer)(conv7),conv2], axis=3) #(120,120,128)
conv8 = Conv2D(64, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(up8) #(120,120,64)
conv8 = Conv2D(64, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(conv8) #(120,120,64)
up9 = concatenate([Conv2DTranspose(32, (2, 2), strides=(2, 2), padding='same', #(240,240,32)
kernel_initializer=initializer)(conv8),conv1], axis=3) #(240,240,64)
conv9 = Conv2D(32, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(up9) #(240,240,32)
conv9 = Conv2D(32, (3, 3), activation='relu', padding='same',
kernel_initializer=initializer)(conv9) #(240,240,32)
conv10 = Conv2D(4, (1, 1), activation='relu',
kernel_initializer=initializer)(conv9) #(240,240,4)
conv10 = Activation('softmax')(conv10) #(240,240,4)
model = tf.keras.Model(inputs = [inputs], outputs = [conv10])
adam = Adam(learning_rate = 1e-4)
model.compile(optimizer= adam, loss = tf.keras.losses.CategoricalCrossentropy(),
metrics=['accuracy'])
return model