python中利用Keras建几层卷积神经网络来做Kaggle猫狗识别

import numpy as np

np.random.seed(1336)  # for reproducibility
from keras.datasets import mnist
from keras.datasets import cifar10
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
import h5py
from PIL import Image
import cv2
import matplotlib.pyplot as pyplot

# 全局变量
batch_size = 20
nb_classes = 2
epochs = 20

trainImagesNum = 4000
testImagesNum = 2000
# input image dimensions
img_rows, img_cols = 128, 128
# number of convolutional filters to use
nb_filters = 128
# size of pooling area for max pooling
pool_size = (2, 2)
# convolution kernel size
kernel_size = (3, 3)
X_train = np.zeros([2000, img_rows, img_cols, 3])
X_test = np.zeros([2000, img_rows, img_cols, 3])

print(type(X_train), X_train.shape)
for i in range(2000):
    if (np.mod(i, 50) == 0):
        print('第', i, '个训练图片正在装载')
    if (i < 1000):
        X_train[i, :, :, :] = cv2.resize(cv2.imread('G:\\train\\cat.' + str(i) + '.jpg'), (img_rows, img_cols))
        # print(np.shape(img),type(img))
    if (i >= 1000):
        X_train[i, :, :, :] = cv2.resize(cv2.imread('G:\\train\\dog.' + str(i - 1000) + '.jpg'), (img_rows, img_cols))

    if (np.mod(i, 50) == 0):
        print('第', i, '个测试图片正在装载')

    X_test[i, :, :, :] = cv2.resize(cv2.imread('G:\\test\\' + str(i + 1) + '.jpg'), (img_rows, img_cols))
    # print(np.shape(img),type(img))

# 训练标签
y_train = np.loadtxt('G:\\trainLabels.txt')
print(y_train[0:20], y_train.shape)
# 测试标签
y_test = np.loadtxt('G:\\testLabels.txt')
print(y_test[0:20], y_test.shape)

index = np.arange(2000)
np.random.shuffle(index)
print(index[0:20])

X_train = X_train[index, :, :, :]
y_train = y_train[index]

##
# t=X_test[999,:,:,:]
# r=Image.fromarray(t[:,:,0]).convert('L')
# g=Image.fromarray(t[:,:,1]).convert('L')
# b=Image.fromarray(t[:,:,2]).convert('L')
# T=Image.merge("RGB",(b,g,r))
# pyplot.imshow(T)
# pyplot.show()


print('标签：', y_train, '形状：', y_train.shape)

input_shape = (img_rows, img_cols, 3)  # 元组类型，元素不能更改
#
X_train = X_train.astype('float32')
X_test = X_test.astype('float32')
y_train = y_train.astype('int32')
y_test = y_test.astype('int32')
print(type(y_train[0]))

X_train /= 255
X_test /= 255
print('X_train shape:', X_train.shape)
print(X_train.shape[0], 'train samples')
print(X_test.shape[0], 'test samples')

print('标签：', y_train, '形状：', y_train.shape)
# 转换为one_hot类型
loc = np.arange(2000)
Y_train = np.zeros((2000, 2))
Y_train[loc, y_train] = 1

#
# Y_test=np.zeros((2000,2))
# Y_test[np.arange(2000),y_test]=1
Y_test = np.zeros((2000, 2))
print(Y_test[0:20])
Y_test[loc, y_test] = 1

# Y_train = np_utils.to_categorical((Y_train), nb_classes)
# Y_test = np_utils.to_categorical((Y_test), nb_classes)

print(Y_train.shape, Y_train[0:20, :])
print(Y_test.shape, Y_test[0:20, :])

# 构建模型
model = Sequential()
""" 
model.add(Convolution2D(nb_filters, kernel_size[0], kernel_size[1], 
                        border_mode='same', 
                        input_shape=input_shape)) 
"""
model.add(Convolution2D(32, (kernel_size[0], kernel_size[1],),
                        padding='same',
                        input_shape=input_shape))  # 卷积层1
model.add(Activation('relu'))  # 激活层
model.add(Convolution2D(32, (kernel_size[0], kernel_size[1]),padding='same'))  # 卷积层2

model.add(Activation('relu'))  # 激活层
model.add(MaxPooling2D(pool_size=pool_size))  # 池化层

model.add(Convolution2D(64, (kernel_size[0], kernel_size[1]),padding='same'))  # 卷积层3
model.add(Activation('relu'))  # 激活层
model.add(Convolution2D(64, (kernel_size[0], kernel_size[1]),padding='same'))  # 卷积层4
model.add(Activation('relu'))  # 激活层
model.add(MaxPooling2D(pool_size=pool_size))  # 池化层

model.add(Convolution2D(128, (kernel_size[0], kernel_size[1]),padding='same'))  # 卷积层3
model.add(Activation('relu'))  # 激活层
model.add(Convolution2D(128, (kernel_size[0], kernel_size[1]),padding='same'))  # 卷积层3
model.add(Activation('relu'))  # 激活层
model.add(MaxPooling2D(pool_size=pool_size))  # 池化层



model.add(Flatten())  # 拉成一维数据
model.add(Dense(1024))  # 全连接层1
model.add(Activation('relu'))  # 激活层
model.add(Dropout(0.5))  # 随机失活
model.add(Dense(nb_classes))  # 全连接层2
model.add(Activation('softmax'))  # Softmax评分

# 编译模型
model.compile(loss='categorical_crossentropy',
              optimizer='adadelta',
              metrics=['accuracy'])
# 训练模型
model.fit(X_train, Y_train, batch_size=batch_size, epochs=epochs,
          verbose=1, validation_data=(X_test, Y_test))
# # 评估模型
score = model.evaluate(X_test, Y_test, verbose=0)
print('Test score:', score[0])
print('Test accuracy:', score[1])