TS05_PIR/CNN_1_cont.py at master · ts05/TS05_PIR · GitHub

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
import Data_load_4 as dl
import os
import numpy as np
import tensorflow as tf
import keras
import Data_load_testdata as dlt
from keras.layers import Input, Dense, Concatenate, Conv2D
from keras.models import Model
from sklearn.model_selection import train_test_split
from tensorflow.python.keras.callbacks import TensorBoard
from keras.callbacks import EarlyStopping
from time import time

current_path = os.getcwd()

DataX = list()
DataY = list()

#Xlength = DataX.__len__() - Xlength
#print(Xlength)

modelname = input("Input model name to save :: ")

totalepoch = 1000

inputPIR = Input(shape=(10, 10, 1))
inputOther = Input(shape=(3,))

model_x1 = Conv2D(32, kernel_size=(3,3), input_shape=(10, 10, 1),  data_format='channels_last', activation='relu', name="Conv1")(inputPIR)
model_x1 = keras.layers.BatchNormalization(name="Batch1")(model_x1)
model_x1 = keras.layers.MaxPooling2D(pool_size=(2,2),name="Max1")(model_x1)
model_x1 = Conv2D(64, kernel_size=(3,3), activation='relu', name="Conv2")(model_x1)
model_x1 = keras.layers.BatchNormalization(name="Batch2")(model_x1)
model_x1 = keras.layers.MaxPooling2D(pool_size=(2,2), name="Max2")(model_x1)
model_x1 = keras.layers.Flatten()(model_x1)

model_x2 = Dense(1, input_shape=(3,))(inputOther)

merged = Concatenate(axis=1)([model_x1, model_x2])

model_x3 = Dense(2, activation='softmax')(merged)
#model_x3 = Dense(2, activation='sigmoid')(merged)

model = Model(inputs=[inputPIR, inputOther], outputs=model_x3)

model.summary()

## Data 순서 :: pir , light, temp , humid
#dl.Data_load("None", DataX, DataY)

#dl.Data_load("Human", DataX, DataY)

dlt.Data_load("None", DataX, DataY)

dlt.Data_load("Human", DataX, DataY)

DataX = np.asarray(DataX)
DataY = np.asarray(DataY)

print(DataX.shape)
print(DataY.shape)

DataX = np.reshape(DataX, (int(DataX.__len__()/4), 4, 100))

x_data = list()
y_data = list()
x_data_test = list()
y_data_test = list()
x_data_other = list()
x_data_other_test = list()

tensorboard = TensorBoard(log_dir="logs/{}".format(time()))

model.compile(optimizer=tf.train.AdamOptimizer(learning_rate=0.0005),
              loss='mean_squared_error',
              metrics=['acc'])

X_train, X_test, Y_train, Y_test = train_test_split(DataX, DataY, test_size=0.3, random_state=777, shuffle=True)

for i in range(len(X_train)):
    x_data.append(X_train[i][0])
    x_data_other.append(X_train[i][1][0])
    x_data_other.append(X_train[i][2][0])
    x_data_other.append(X_train[i][3][0])

for i in range(len(X_test)):
    x_data_test.append(X_test[i][0])
    x_data_other_test.append(X_test[i][1][0])
    x_data_other_test.append(X_test[i][2][0])
    x_data_other_test.append(X_test[i][3][0])

x_data = np.asarray(x_data)
x_data_other = np.asarray(x_data_other)

x_data = np.reshape(x_data, (-1, 10, 10, 1))
x_data_other = np.reshape(x_data_other, (-1, 3))
x_data_test = np.reshape(x_data_test, (-1, 10, 10, 1))
x_data_other_test = np.reshape(x_data_other_test, (-1,3))

k = 0
for i in range(0, DataY.__len__()):
    if DataY[i][0] == 1:
        k = k + 1

print("None 데이터셋 갯수 :: " + str(k))

print(X_train.shape)
print(Y_train.shape)

print(x_data.shape)
print(x_data_other.shape)
print(x_data_test.shape)

#early_stopping = EarlyStopping(monitor='val_loss', min_delta=0, patience=12, verbose=0, mode='auto')
model.fit([x_data,x_data_other], Y_train, epochs=totalepoch, batch_size=256, shuffle='True', validation_data=([x_data_test,x_data_other_test], Y_test), verbose=1, callbacks=[tensorboard])

x_validate = x_data[:3000]
x_validate_other = x_data_other[:3000]
y_validate = Y_train[:3000]
print(y_validate)

results = model.evaluate([x_validate,x_validate_other], y_validate)

print("Validate data[loss, accuracy] :: ")
print(results)

#print(model.predict_classes(X_test[:1, :], verbose=0))
#print('----------------------------------------------')

model_json = model.to_json()

with open("Models/" + modelname + ".json", "w") as json_file :
    json_file.write(model_json)

model.save_weights("Models/" + modelname + ".h5")

print(modelname)
print("Model Saved...!")
print("Train Dataset .. :: ")
print(X_train.shape)
print("Validate Dataset .. :: ")
print(X_test.shape)