VGG+ResNet(Fashion_MNIST)

앞서 간단한 CNN 구조를 이용하여 FASHION-MNIST data를 학습을 시켰었다.   (참고)

keras는 Sequential model, Functional API을 사용할 수 있는데,

간단하게 모델을 구성할때는 Sequential model로 조금 복잡한 모델은 Functional API을 이용하여 model을 만들수 있습니다.

이번에는 Keras의 Functional API이용하여 복잡한 구조의 모델을 한번 짜보도록 하겠습니다.

                                 

                                    [VGG16 model]                                                [ResNet model]

그림 출처 :cs213n [link]

그렇다면 이 두 모델의 핵심을 Keras의 Functional API을 이용하여 핵심 부분을 합쳐보면 어떨까?

[합친 Model의 구조] 

[Fashion_MNIST data Training]

성능으로는 약 88%정도 나오는것을 확인할 수 있었으며, 기존 구조가 간단한 CNN구성을 통하여 학습을 했을경우 보다 안좋은 성능이 나왔으며, 구조가 복잡하거나 깊다고 다 학습이 잘안되는 모습을 확인할 수 있었습니다.

import numpy as np
import mnist_reader
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense
from keras.layers import Dropout
from keras.layers import Flatten
from keras.layers.convolutional import Convolution2D
from keras.layers.convolutional import MaxPooling2D
from keras.utils import np_utils
from matplotlib import pyplot as plt
import time
from keras.layers.normalization import BatchNormalization
from keras.layers import Conv2D, Input, Activation
from keras.models import Model
from keras.layers.convolutional import Convolution2D, MaxPooling2D, ZeroPadding2D
import keras
from keras.layers import merge
 
#Time
start_time = time.time()
 
 
#data load
x_train, y_train = mnist_reader.load_mnist('data/', kind='train')
x_test, y_test = mnist_reader.load_mnist('data/', kind='t10k')
x_train = x_train.reshape(x_train.shape[0], 28, 28, 1).astype('float32')
x_test = x_test.reshape(x_test.shape[0], 28, 28, 1).astype('float32')
 
# normalize inputs from 0-255 to 0-1
x_train = x_train / 255
x_test = x_test / 255
 
# one hot encode outputs
y_train = np_utils.to_categorical(y_train)
y_test = np_utils.to_categorical(y_test)
num_classes = y_test.shape[1]

 
 
# create model
input_img = Input(shape=(28, 28, 1), name='main_input')
 
 
 
#VGG Net
x1 = Conv2D(64, (3, 3))(input_img)
x1 = Activation('relu')(x1)
x1 = Conv2D(64, (3, 3))(x1)
x1 = Activation('relu')(x1)
x1 = MaxPooling2D()(x1)
x1 = Conv2D(64, (3, 3))(x1)
x1 = Activation('relu')(x1)
x1 = Conv2D(64, (3, 3))(x1)
x1 = Activation('relu')(x1)
x1 = MaxPooling2D()(x1)
x1 = Conv2D(64, (3, 3))(x1)
x1 = Activation('relu')(x1)
x1 = MaxPooling2D()(x1)
x1 = Flatten()(x1)
x1 = Dense(256)(x1)
x1 = BatchNormalization()(x1)
x1 = Activation('relu')(x1)
x1 = Dense(256)(x1)
x1 = BatchNormalization()(x1)
x1 = Activation('relu')(x1)
 
 
#Res Net
x = Conv2D(64, (3, 3))(input_img)
x = BatchNormalization()(x)
x = Activation('relu')(x)
x = (ZeroPadding2D((1,1)))(x)
x = Conv2D(64, (3, 3))(input_img)
x = BatchNormalization()(x)
x = Activation('relu')(x)
x = Conv2D(1, (3, 3))(input_img)
x = BatchNormalization()(x)
x = Activation('relu')(x)
x = (ZeroPadding2D((1,1)))(x)
x = merge([x, input_img], mode='sum')
x = Flatten()(x)
x = Dense(256)(x)
x = BatchNormalization()(x)
x = Activation('relu')(x)
x = Dense(256)(x)
x = BatchNormalization()(x)
x = Activation('relu')(x)
 
x = keras.layers.concatenate([x1, x])
out = Dense(num_classes, activation='softmax')(x)
 
# Compile model
model = Model(inputs=input_img, outputs=out)
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
 
#model 가시화 만들기
#from IPython.display import SVG
#from keras.utils.vis_utils import model_to_dot
#SVG(model_to_dot(model, show_shapes=True).create(prog='dot', format='svg'))
 
 
 
# Fit the model
hist = model.fit(x_train, y_train, validation_data=(x_test, y_test), nb_epoch=10, batch_size=50, verbose=2)
# Final evaluation of the model
scores = model.evaluate(x_test, y_test, verbose=0)
print("CNN Error: %.2f%%" % (100-scores[1]*100))
 
 
#모델 시각
fig, loss_ax = plt.subplots()
 
acc_ax = loss_ax.twinx()
 
loss_ax.plot(hist.history['loss'], 'y', label='train loss')
loss_ax.plot(hist.history['val_loss'], 'r', label='val loss')
 
acc_ax.plot(hist.history['acc'], 'b', label='train acc')
acc_ax.plot(hist.history['val_acc'], 'g', label='val acc')
 
loss_ax.set_xlabel('epoch')
loss_ax.set_ylabel('loss')
acc_ax. set_ylabel('accuracy')
 
loss_ax.legend(loc='upper left')
acc_ax.legend(loc='lower left')
 
plt.show()
 
#End Time
print("--- %s seconds ---" %(time.time() - start_time))
 

[전체 Code]