forked from AdityaSoni19031997/Machine-Learning
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdogs_cat_2.py
238 lines (214 loc) · 9.96 KB
/
dogs_cat_2.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
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
# Choose the underlying compiler - tensorflow or theano
import json
import os
with open(os.path.expanduser('~') + "/.keras/keras.json","r") as f:
compiler_data = json.load(f)
compiler_data["backend"] = "tensorflow"
compiler_data["image_data_format"] = "channels_last"
with open(os.path.expanduser('~') + '/.keras/keras.json', 'w') as outfile:
json.dump(compiler_data, outfile)
# import all the required packages
import numpy as np
from keras.models import Model
import keras.backend as K
import keras
from keras import layers
from keras import metrics
from keras.layers import Dense, Convolution2D, BatchNormalization, Dropout, Flatten, AveragePooling2D, Activation, Input, MaxPooling2D
from keras.optimizers import SGD
if False:
from keras.engine.topology import Merge
defined_metrics = []
defined_loss = ''
# Load data from folder images
import os
import PIL
from PIL import Image
import numpy as np
from keras.utils import np_utils
train_data = []
train_label = []
train_label_names = []
trainRootPath = '/Path_to_traindataset'
allFolders = os.listdir(trainRootPath)
for i, cfolder in enumerate(allFolders):
allFiles = os.listdir(trainRootPath + cfolder)
for j, cimage in enumerate(allFiles):
img = Image.open( trainRootPath + cfolder + '/' + cimage ).convert('LA')
img = img.resize((299,299), PIL.Image.ANTIALIAS)
img.load()
cdata = np.asarray( img, dtype="int32" )
train_data.append(cdata)
train_label.append(i)
train_label_names.append(cfolder)
train_data = np.array(train_data)
train_label = np.array(train_label)
train_label = np_utils.to_categorical(train_label, len(set(train_label)))
val_data = []
val_label = []
val_label_names = []
if('/Path_to_valdataset'):
valRootPath = '/Path_to_valdataset'
allFolders = os.listdir(valRootPath)
for i, cfolder in enumerate(allFolders):
allFiles = os.listdir(valRootPath + cfolder)
for j, cimage in enumerate(allFiles):
img = Image.open( valRootPath + cfolder + '/' + cimage ).convert('LA')
img = img.resize((299,299), PIL.Image.ANTIALIAS)
img.load()
cdata = np.asarray( img, dtype="int32" )
val_data.append(cdata)
val_label.append(i)
val_label_names.append(cfolder)
val_data = np.array(val_data)
val_label = np.array(val_label)
val_label = np_utils.to_categorical(val_label, len(set(val_label)))
test_data = []
test_label = []
test_label_names = []
if('/Path_to_testdataset'):
testRootPath = '/Path_to_testdataset'
allFolders = os.listdir(testRootPath)
for i, cfolder in enumerate(allFolders):
allFiles = os.listdir(testRootPath + cfolder)
for j, cimage in enumerate(allFiles):
img = Image.open( testRootPath + cfolder + '/' + cimage ).convert('LA')
img = img.resize((299,299), PIL.Image.ANTIALIAS)
img.load()
cdata = np.asarray( img, dtype="int32" )
test_data.append(cdata)
test_label.append(i)
test_label_names.append(cfolder)
test_data = np.array(test_data)
test_label = np.array(test_label)
test_label = np_utils.to_categorical(test_label, len(set(test_label)))
print(train_data.shape)
batch_input_shape_Custom = train_data.shape[1:]
train_batch_size = 16
# Choose the hardware platform - GPU or CPU
import tensorflow as tf
if('cpu' in 'CPU'.lower()):
device_id = '/cpu'
if('gpu' in 'CPU'.lower()):
device_id = '/gpu:0'
if('multigpu' in 'CPU'.lower()):
device_id = '/gpu:1'
with tf.device(device_id):
#Input Layer
Custom_0 = Input(shape=batch_input_shape_Custom)
#Convolution2D Layer
Convolution2D_1 = Convolution2D(64, 3, 3, init = 'lecun_uniform', border_mode = 'valid', subsample = (1, 1), dim_ordering = 'tf', bias = False, name = 'Convolution2D.c18db6db')(Custom_0)
#Batch Normalization Layer
if K.image_dim_ordering() == 'th':
bn_axis = 1
else:
bn_axis = 3
Convolution2D_1 = BatchNormalization(axis=bn_axis,name='bn_Convolution2D.c18db6db')(Convolution2D_1)
#Rectification Linear Unit (ReLU) Activation Layer
ReLU_2 = Activation('relu', name = 'ReLU.7e9f24ad')(Convolution2D_1)
#Dropout Layer
Dropout_3 = Dropout(0.35, name = 'Dropout.69d8f55e')(ReLU_2)
#Convolution2D Layer
Convolution2D_4 = Convolution2D(64, 3, 3, init = 'lecun_uniform', border_mode = 'valid', subsample = (1, 1), dim_ordering = 'tf', bias = False, name = 'Convolution2D.3a9ebc96')(Dropout_3)
#Batch Normalization Layer
if K.image_dim_ordering() == 'th':
bn_axis = 1
else:
bn_axis = 3
Convolution2D_4 = BatchNormalization(axis=bn_axis,name='bn_Convolution2D.3a9ebc96')(Convolution2D_4)
#Rectification Linear Unit (ReLU) Activation Layer
ReLU_5 = Activation('relu', name = 'ReLU.bda49317')(Convolution2D_4)
#Pooling2D Layer
Pooling2D_6 = MaxPooling2D(pool_size = (3, 3), border_mode = 'valid', strides = (3, 3), name = 'Pooling2D.79b7880e')(ReLU_5)
#Pooling2D Layer
Pooling2D_7 = MaxPooling2D(pool_size = (3, 3), border_mode = 'valid', strides = (3, 3), name = 'Pooling2D.e4e8b64e')(Pooling2D_6)
#Convolution2D Layer
Convolution2D_8 = Convolution2D(128, 3, 3, init = 'lecun_uniform', border_mode = 'valid', subsample = (1, 1), dim_ordering = 'tf', bias = False, name = 'Convolution2D.2174a78f')(Pooling2D_7)
#Batch Normalization Layer
if K.image_dim_ordering() == 'th':
bn_axis = 1
else:
bn_axis = 3
Convolution2D_8 = BatchNormalization(axis=bn_axis,name='bn_Convolution2D.2174a78f')(Convolution2D_8)
#Rectification Linear Unit (ReLU) Activation Layer
ReLU_9 = Activation('relu', name = 'ReLU.ae7c0ba5')(Convolution2D_8)
#Dropout Layer
Dropout_10 = Dropout(0.35, name = 'Dropout.c100b76a')(ReLU_9)
#Convolution2D Layer
Convolution2D_11 = Convolution2D(128, 3, 3, init = 'lecun_uniform', border_mode = 'valid', subsample = (1, 1), dim_ordering = 'tf', bias = False, name = 'Convolution2D.ca77e78d')(Dropout_10)
#Batch Normalization Layer
if K.image_dim_ordering() == 'th':
bn_axis = 1
else:
bn_axis = 3
Convolution2D_11 = BatchNormalization(axis=bn_axis,name='bn_Convolution2D.ca77e78d')(Convolution2D_11)
#Rectification Linear Unit (ReLU) Activation Layer
ReLU_12 = Activation('relu', name = 'ReLU.6011e9ac')(Convolution2D_11)
#Pooling2D Layer
Pooling2D_13 = MaxPooling2D(pool_size = (3, 3), border_mode = 'valid', strides = (3, 3), name = 'Pooling2D.68bcd3de')(ReLU_12)
#Pooling2D Layer
Pooling2D_14 = MaxPooling2D(pool_size = (3, 3), border_mode = 'valid', strides = (3, 3), name = 'Pooling2D.e659d7ac')(Pooling2D_13)
#Convolution2D Layer
Convolution2D_15 = Convolution2D(256, 3, 3, init = 'lecun_uniform', border_mode = 'valid', subsample = (1, 1), dim_ordering = 'tf', bias = False, name = 'Convolution2D.a4814e4')(Pooling2D_14)
#Batch Normalization Layer
if K.image_dim_ordering() == 'th':
bn_axis = 1
else:
bn_axis = 3
Convolution2D_15 = BatchNormalization(axis=bn_axis,name='bn_Convolution2D.a4814e4')(Convolution2D_15)
#Rectification Linear Unit (ReLU) Activation Layer
ReLU_16 = Activation('relu', name = 'ReLU.a70b041')(Convolution2D_15)
#Dropout Layer
Dropout_17 = Dropout(0.35, name = 'Dropout.db88dc2c')(ReLU_16)
#Convolution2D Layer
Convolution2D_18 = Convolution2D(256, 3, 3, init = 'lecun_uniform', border_mode = 'valid', subsample = (1, 1), dim_ordering = 'tf', bias = False, name = 'Convolution2D.d2e72e06')(Dropout_17)
#Batch Normalization Layer
if K.image_dim_ordering() == 'th':
bn_axis = 1
else:
bn_axis = 3
Convolution2D_18 = BatchNormalization(axis=bn_axis,name='bn_Convolution2D.d2e72e06')(Convolution2D_18)
#Rectification Linear Unit (ReLU) Activation Layer
ReLU_19 = Activation('relu', name = 'ReLU.45fb4db9')(Convolution2D_18)
#Pooling2D Layer
Pooling2D_20 = MaxPooling2D(pool_size = (3, 3), border_mode = 'valid', strides = (3, 3), name = 'Pooling2D.a3ec2ea8')(ReLU_19)
#Pooling2D Layer
Pooling2D_21 = MaxPooling2D(pool_size = (3, 3), border_mode = 'valid', strides = (3, 3), name = 'Pooling2D.969f8798')(Pooling2D_20)
#Flatten Layer
Flatten_22 = Flatten(name = 'Flatten.497d859e')(Pooling2D_21)
#Dense or Fully Connected (FC) Layer
Dense_23 = Dense(1024, init = 'lecun_uniform', bias = True, name = 'Dense.4fac24ec')(Flatten_22)
#Sigmoid Activation Layer
Sigmoid_24 = Activation('sigmoid', name = 'Sigmoid.5cde8524')(Dense_23)
#Dropout Layer
Dropout_25 = Dropout(0.5, name = 'Dropout.151333e6')(Sigmoid_24)
#Dense or Fully Connected (FC) Layer
Dense_26 = Dense(512, init = 'lecun_uniform', bias = True, name = 'Dense.fe94335a')(Dropout_25)
#Sigmoid Activation Layer
Sigmoid_27 = Activation('sigmoid', name = 'Sigmoid.ace05b6e')(Dense_26)
#Dropout Layer
Dropout_28 = Dropout(.3, name = 'Dropout.e7f5e01d')(Sigmoid_27)
#Dense or Fully Connected (FC) Layer
Dense_29 = Dense(2, init = 'uniform', bias = True, name = 'Dense.eee2cd46')(Dropout_28)
#Softmax Activation Layer
SoftmaxWithLoss_30 = Activation('softmax', name = 'SoftmaxWithLoss.2dad1fd6')(Dense_29)
defined_loss = 'binary_crossentropy'
# Define a keras model
model = Model(input=[Custom_0, SoftmaxWithLoss_30], output = [])
# Set the required hyperparameters
num_epochs = 100
# Defining the optimizer function
optimizer_fn = SGD(lr=0.01, momentum=0.9, decay=0.8)
# Compile and train the model
if not defined_metrics:
defined_metrics=None
if not defined_loss:
defined_loss = 'categorical_crossentropy'
model.compile(loss=defined_loss, optimizer=optimizer_fn, metrics=defined_metrics)
model.fit(train_data, train_label, batch_size=train_batch_size, nb_epoch=num_epochs, verbose=1)
# validate the model
if (val_data):
val_scores = model.evaluate(val_data, val_label, verbose=1)
# test the model
if (test_data):
test_scores = model.evaluate(test_data, test_label, verbose=1)