-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy patharchitecture.py
128 lines (104 loc) · 3.62 KB
/
architecture.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
from keras.models import Sequential
from keras.layers import *
from skimage.measure import compare_psnr
import numpy as np
import keras.backend as K
from keras.callbacks import *
def psnr(base_image,altered_image):
try:
MSE=K.mean(K.square(base_image-altered_image))
if(MSE==0):
return 100
else:
return 20*K.log(255.0/K.sqrt(MSE))
except Exception as e:
print(e)
return K.constant(100)
def advance_relu(input_):
return K.relu( input_,alpha=0.0, max_value=255.0)
import pickle
bth=50
epc=1
model=Sequential()
mc=ModelCheckpoint("weights.{epoch:02d}-loss{loss:.2f}-acc{acc:.2f}-.hdf5", monitor='loss', verbose=1, save_best_only=False, save_weights_only=False, mode='auto', period=1)
es=EarlyStopping(monitor='loss', min_delta=1, patience=3, verbose=1, mode='auto')
tb=TensorBoard(log_dir='./logs', batch_size=bth, write_graph=True, write_grads=True, write_images=True, embeddings_freq=0, embeddings_layer_names=None, embeddings_metadata=None)
lr=ReduceLROnPlateau(monitor='loss', factor=0.1, patience=1, verbose=1, mode='auto', epsilon=0.0001, cooldown=0, min_lr=0)
model.add(Conv2D(64,(3,3),input_shape=(256,256,3),strides=(1,1),padding="valid",
activation="linear",
kernel_initializer="truncated_normal",
bias_initializer="truncated_normal"))
model.add(BatchNormalization())
model.add(Activation("relu"))
model.add(MaxPooling2D(pool_size=(2,2)))
#conv2_1
model.add(Conv2D(128,(2,2),strides=(1,1),padding="valid",
activation="linear",
kernel_initializer="truncated_normal",
bias_initializer="truncated_normal"))
model.add(BatchNormalization())
model.add(Activation("relu"))
#conv2_2
model.add(MaxPooling2D(pool_size=(2,2)))
#conv3_1
model.add(Conv2D(256,(2,2),strides=(1,1),padding="valid",
activation="linear",
kernel_initializer="truncated_normal",
bias_initializer="truncated_normal"))
model.add(BatchNormalization())
model.add(Activation("relu"))
#conv3_2
model.add(MaxPooling2D(pool_size=(2,2)))
#CONV4_1
model.add(Conv2D(512,(2,2),strides=(1,1),padding="valid",
activation="linear",
kernel_initializer="truncated_normal",
bias_initializer="truncated_normal"))
model.add(BatchNormalization())
model.add(Activation("relu"))
model.add(Dropout(0.5))
########################################
## DECONV STARTS
########################################
#DECONV 1 -1
model.add(Conv2DTranspose(512,(2,2),strides=(1,1),padding="valid",
activation="linear",
kernel_initializer="truncated_normal",
bias_initializer="truncated_normal"))
model.add(BatchNormalization())
model.add(Activation("relu"))
model.add(UpSampling2D(size=(2,2)))
# DECONV 2-1
model.add(Conv2DTranspose(256,(2,2),strides=(1,1),padding="valid",
activation="linear",
kernel_initializer="truncated_normal",
bias_initializer="truncated_normal"))
model.add(BatchNormalization())
model.add(Activation("relu"))
model.add(UpSampling2D(size=(2,2)))
#DECONV 3-1
model.add(Conv2DTranspose(128,(2,2),strides=(1,1),padding="valid",
activation="linear",
kernel_initializer="truncated_normal",
bias_initializer="truncated_normal"))
model.add(BatchNormalization())
model.add(Activation("relu"))
model.add(UpSampling2D(size=(2,2)))
# DECONV 4-1
model.add(Conv2DTranspose(3,(3,3),strides=(1,1),padding="valid",
activation=advance_relu,
kernel_initializer="truncated_normal",
bias_initializer="truncated_normal"))
model.summary()
k=model.get_config()
print(k)
print("values")
print(k.values)
model.compile(
loss="mse",
optimizer="adam",
metrics=['accuracy',psnr]
)
with open("./data.pkl",'rb') as f:
data=pickle.load(f)
model.fit(data['norain'].reshape(-1,256,256,3)[:100,:],data['rain'].reshape(-1,256,256,3)[:100,:],epochs=epc,batch_size=bth,shuffle=True,validation_split=0.3,callbacks=[mc,es,tb,lr])