from keras.applications.vgg16 import VGG16
from keras.models import Sequential
from keras.layers import Conv2D,MaxPool2D,Activation,Dropout,Flatten,Dense
from keras.optimizers import SGD
from keras.preprocessing.image import ImageDataGenerator,img_to_array,load_img
import numpy as np

1 vgg16_model = VGG16(weights='imagenet',include_top=False, input_shape=(150,150,3))

 1 # 搭建全连接层
 2 top_model = Sequential()
 3 top_model.add(Flatten(input_shape=vgg16_model.output_shape[1:]))
 4 top_model.add(Dense(256,activation='relu'))
 5 top_model.add(Dropout(0.5))
 6 top_model.add(Dense(2,activation='softmax'))
 7 
 8 model = Sequential()
 9 model.add(vgg16_model)
10 model.add(top_model)

train_datagen = ImageDataGenerator(
    rotation_range = 40,     # 随机旋转度数
    width_shift_range = 0.2, # 随机水平平移
    height_shift_range = 0.2,# 随机竖直平移
    rescale = 1/255,         # 数据归一化
    shear_range = 20,       # 随机错切变换
    zoom_range = 0.2,        # 随机放大
    horizontal_flip = True,  # 水平翻转
    fill_mode = 'nearest',   # 填充方式
) 
test_datagen = ImageDataGenerator(
    rescale = 1/255,         # 数据归一化
)

batch_size = 32

# 生成训练数据
train_generator = train_datagen.flow_from_directory(
    'image/train',
    target_size=(150,150),
    batch_size=batch_size,
    )

# 测试数据
test_generator = test_datagen.flow_from_directory(
    'image/test',
    target_size=(150,150),
    batch_size=batch_size,
    )

train_generator.class_indices

{'cat': 0, 'dog': 1}

1 # 定义优化器，代价函数，训练过程中计算准确率
2 model.compile(optimizer=SGD(lr=1e-4,momentum=0.9),loss='categorical_crossentropy',metrics=['accuracy'])
3 
4 model.fit_generator(train_generator,steps_per_epoch=len(train_generator),epochs=20,validation_data=test_generator,validation_steps=len(test_generator))

# pip install h5py
model.save('model_vgg16.h5')

测试

from keras.models import load_model
import numpy as np

label = np.array(['cat','dog'])
# 载入模型
model = load_model('model_vgg16.h5')

# 导入图片
image = load_img('image/test/cat/cat.1003.jpg')
image

image = image.resize((150,150))
image = img_to_array(image)
image = image/255
image = np.expand_dims(image,0)
image.shape

(1, 150, 150, 3)

print(label[model.predict_classes(image)]

['cat']

使用VGG16完成猫狗分类

猜你喜欢