/
cnn_capstone.py
200 lines (178 loc) · 7.64 KB
/
cnn_capstone.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
'''Trains a simple convnet on the initial flower dataset.
Adapted cnn_soln.py, which was
from https://github.com/fchollet/keras/blob/master/examples/mnist_cnn.py
'''
from __future__ import print_function
import numpy as np
import os
from os import listdir
from os.path import isfile, join
import re
import PIL
from PIL import Image
from skimage import io
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelEncoder
np.random.seed(1337) # for reproducibility
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation, Flatten
from keras.layers import Convolution2D, MaxPooling2D
from keras.utils import np_utils
from keras import backend as K
# def load_data(directory_path):
# for filename in os.listdir(directory_path):
#append filedata to something like a dictionary or mongodb
#return dictionary or mongodb
# def my_list_dir(root):
# filelist = os.listdir(root)
# return [x for x in filelist if not (x.startswith('.'))]
# def clean_file_paths(root):
# for filename in my_list_dir(root):
# print(filename)
# # print(path, subdirs, files)
# name2 = filename.replace(' ', '')
# new_name = name2.replace('-', '_')
# os.replace(filename, new_name)
def my_image_resize(basewidth, root, resized_root):
files = [f for f in listdir(root) if isfile(join(root, f))]
resized_files = [f for f in listdir(resized_root) if isfile(join(resized_root, f))]
# os.mkdir('../resized_images')
# for path, subdirs, files in os.walk(root):
for name in files:
if not (name.startswith('.')):
if not ('{}_resized.png'.format(name[:-4])) in resized_files:
# if name != 'cnn_capstone.py':
img = Image.open('{}{}'.format(root, name))
wpercent = (basewidth / float(img.size[0]))
hsize = int((float(img.size[1]) * float(wpercent)))
img = img.resize((basewidth, hsize), PIL.Image.ANTIALIAS)
img.save('../resized_images/{}_resized.png'.format(name[:-4]))
def image_categories(resized_root):
''' A dictionary that stores the image path name and flower species for each image
Input: image path names (from root directory)
Output: dictionary 'categories'
'''
flower_dict = {}
files = [f for f in listdir(resized_root) if isfile(join(resized_root, f))]
# for path, subdirs, files in os.walk(resized_root):
# print(path, subdirs, files)
for name in files:
# name = name.replace(' ', '')
# name = name.replace('-', '_')
# if not (name.startswith('.')):
# if name != 'cnn_capstone.py':
img_path = '{}/{}'.format(resized_root, name)
# img_path = os.path.join(path, name)
img_cat = re.sub("\d+", "", name).rstrip('_resized.png')
img_cat = img_cat[:-3]
flower_dict[img_path] = img_cat
return flower_dict
def my_train_test_split(resized_root):
X = []
y = []
files = [f for f in listdir(resized_root) if isfile(join(resized_root, f))]
# files = [f for f in os.listdir(os.curdir) if os.path.isfile(f)]
# resized_files = [f for f in files if f[-11:] == 'resized.png']
# for path, subdirs, files in os.walk(root):
for name in files:
if not (name.startswith('.')):
# if name != 'cnn_capstone.py':
# name = name.replace(' ', '')
# name = name.replace('-', '_')
img_path = '{}/{}'.format(resized_root, name)
# img_path = os.path.join(path, name)
correct_cat = categories[img_path]
img_pixels = io.imread('{}/{}'.format(resized_root, name))
X.append(img_pixels)
y.append(correct_cat)
y = np.array(y)
number = LabelEncoder()
y = number.fit_transform(y.astype('str'))
X_arr = np.array(X)
# X_arr = np.stack(X)
X_train, X_test, y_train, y_test = train_test_split(X_arr, y)
return X_train, X_test, y_train, y_test
# def image_reshape(X_train, X_test):
# if K.image_dim_ordering() == 'th':
# X_train = X_train.reshape(X_train.shape[0], 1, img_rows, img_cols)
# X_test = X_test.reshape(X_test.shape[0], 1, img_rows, img_cols)
# input_shape = (1, img_rows, img_cols)
# else:
# X_train = X_train.reshape(X_train.shape[0], img_rows, img_cols, 1)
# X_test = X_test.reshape(X_test.shape[0], img_rows, img_cols, 1)
# input_shape = (img_rows, img_cols, 1)
# return X_train, X_test, input_shape
def image_asfloat(X_train, X_test):
X_train = X_train.astype('float32')
X_test = X_test.astype('float32')
return X_train, X_test
def image_rgb_unit_scale(X_train, X_test):
X_train /= 255
X_test /= 255
return X_train, X_test
def print_X_shapes(X_train, X_test):
print('X_train shape:', X_train.shape)
print(X_train.shape[0], 'train samples')
print(X_test.shape[0], 'test samples')
def convert_to_binary_class_matrices(y_train, y_test, nb_classes):
# convert class vectors to binary class matrices
Y_train = np_utils.to_categorical(y_train, nb_classes)
Y_test = np_utils.to_categorical(y_test, nb_classes)
return Y_train, Y_test
def cnn_model(nb_filters, kernel_size, batch_size, nb_epoch, X_test, Y_test, X_train, Y_train):
model = Sequential()
model.add(Convolution2D(nb_filters, (kernel_size[0], kernel_size[1]),
border_mode='valid',
input_shape=(img_rows, img_cols, 3)))
model.add(Activation('relu'))
model.add(Convolution2D(nb_filters, (kernel_size[0], kernel_size[1])))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=pool_size))
model.add(Dropout(0.25))
# model.add(Convolution2D(nb_filters, (kernel_size[0], kernel_size[1])))
# model.add(Activation('relu'))
# model.add(Convolution2D(nb_filters, (kernel_size[0], kernel_size[1])))
# model.add(Activation('relu'))
model.add(Flatten())
model.add(Dense(128))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(nb_classes))
model.add(Activation('softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='adadelta',
metrics=['accuracy'])
model.fit(X_train, Y_train, batch_size=batch_size, epochs=nb_epoch,
verbose=1, validation_data=(X_test, Y_test))
score = model.evaluate(X_test, Y_test, verbose=0)
ypred = model.predict(X_test)
print('Test score:', score[0])
print('Test accuracy:', score[1])
return ypred
if __name__ == '__main__':
# filelist = my_list_dir('../data_images/')
# clean_file_paths('../data_images/')
batch_size = 5
# number of flowers in dataset
nb_classes = 11
#number of repeats of entire model
nb_epoch = 20
# input image dimensions
img_rows, img_cols = 120, 90
# number of convolutional filters to use
nb_filters = 32
# size of pooling area for max pooling
pool_size = (2, 2)
# convolution kernel size
kernel_size = (3, 3)
root = '../data_images/'
resized_root = '../resized_images'
# my_image_resize(90, '../data_images/', resized_root)
categories = image_categories(resized_root)
X_train, X_test, y_train, y_test = my_train_test_split(resized_root)
X_train, X_test = image_asfloat(X_train, X_test)
X_train, X_test = image_rgb_unit_scale(X_train, X_test)
print_X_shapes(X_train, X_test)
Y_train, Y_test = convert_to_binary_class_matrices(y_train, y_test, nb_classes)
ypred = cnn_model(nb_filters, kernel_size, batch_size, nb_epoch, X_test, Y_test, X_train, Y_train)
# ypred = cnn_model(nb_filters, kernel_size, batch_size, nb_epoch, X_test, Y_test, X_train, Y_train)