吴裕雄--天生自然神经网络与深度学习实战Python+Keras+TensorFlow：构造一个识别猫、狗图片的卷积网络

import os
base_dir = '/Users/apple/Documents/cat-and-dog'
train_cats_dir = os.path.join(base_dir, 'training_set/cats')
train_dogs_dir = os.path.join(base_dir, 'training_set/dogs')
test_cats_dir = os.path.join(base_dir, 'test_set/cats')
test_dogs_dir = os.path.join(base_dir, 'test_set/dogs')

print('total trainning cat images: ', len(os.listdir(train_cats_dir)))
print('total trainning dog images: ', len(os.listdir(train_dogs_dir)))
print('total testing cat images: ', len(os.listdir(test_cats_dir)))
print('total testing cat images: ', len(os.listdir(test_dogs_dir)))

model = models.Sequential()
model.add(layers.Conv2D(32, (3,3),  activation='relu', input_shape=(150,150,3)))
model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Conv2D(64, (3,3), activation='relu'))
model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Conv2D(128, (3,3), activation='relu'))
model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Conv2D(128, (3,3), activation='relu'))
model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Flatten())
model.add(layers.Dense(512, activation='relu'))
model.add(layers.Dense(1, activation="sigmoid"))
model.compile(loss='binary_crossentropy', optimizer='rmsprop', metrics=['acc'])
model.summary()

from keras import preprocessing
from keras.preprocessing import image

import numpy as np 

path_cats = []
for path in os.listdir(train_cats_dir):
    if '.jpg' in path:
        path_cats.append(os.path.join(train_cats_dir, path))
path_dogs = []
for path in os.listdir(train_dogs_dir):
    if '.jpg' in path:
        path_dogs.append(os.path.join(train_dogs_dir, path))


training_set = np.zeros((6000, 150, 150, 3), dtype='float32')
train_dog_imgs = 3000
train_cat_imgs = 3000
for i in range(0, train_dog_imgs):
    img = preprocessing.image.load_img(path_dogs[i], target_size=(150,150))
    training_set[i] = preprocessing.image.img_to_array(img)
for j in range(0, train_cat_imgs):
    img = preprocessing.image.load_img(path_cats[j], target_size=(150,150))
    training_set[train_dog_imgs+j] = preprocessing.image.img_to_array(img)


validation_set = np.zeros((2000, 150, 150, 3), dtype='float32')
validation_dog_imgs = 1000
validation_cat_imgs = 1000
for i in range(validation_dog_imgs):
    path = path_dogs[i + train_dog_imgs] 
    img = preprocessing.image.load_img(path, target_size=(150,150))
    validation_set[i] = preprocessing.image.img_to_array(img)
for j in range(validation_cat_imgs):
    path = path_cats[j+train_cat_imgs]
    img = preprocessing.image.load_img(path, target_size=(150,150))
    validation_set[j+validation_dog_imgs] = preprocessing.image.img_to_array(img)

train_labels = np.zeros((3000,))
#猫图片打标签1
train_labels = np.concatenate((train_labels, np.ones(3000, )))
validation_labels = np.zeros((1000, ))
validation_labels = np.concatenate((validation_labels, np.ones(1000, )))

train_datagen = preprocessing.image.ImageDataGenerator(rescale=1./255)
train_generator = train_datagen.flow(training_set, train_labels, 
                                         batch_size = 32)
validation_generator = train_datagen.flow(validation_set, validation_labels,
                                         batch_size=32)

history = model.fit_generator(train_generator, steps_per_epoch=100,
                             epochs = 30, validation_steps=50,
                             validation_data=validation_generator)

model.save('cats_and_dogs_small_1.h5')

import matplotlib.pyplot as plt

acc = history.history['acc']
val_acc = history.history['val_acc']
loss = history.history['loss']
val_loss = history.history['val_loss']

epochs = range(1, len(acc) + 1)

plt.plot(epochs, acc, 'bo', label='Training acc')
plt.plot(epochs, val_acc, 'b', label='Validation acc')
plt.title('Training and validation accuracy')
plt.legend()
plt.figure()
plt.plot(epochs, loss, 'bo', label='Training loss')
plt.plot(epochs, val_loss, 'b', label='Validation loss')
plt.title('Training and validation loss')
plt.legend()
plt.show()