猫狗大战 ---使用预训练模型之微调

0X00 概念

模型微调指对于用特征提取的冻结的模型基，将模型顶部的几层“解冻”，并将解冻的几层和新增加的几层联合训练。

0X01 导入模型

from keras.applications import VGG16

conv_base = VGG16(weights='imagenet',
                 include_top=False,
                 input_shape=(150, 150, 3)
                 )
#查看模型
conv_base.summary()

0x02 解冻知道某一层的所有层

conv_base.trainable = True

set_trainable = False
for layer in conv_base.layers:
    if layer.name == 'block5_conv1 (Conv2D) ':
        set_trainable = True
    if set_trainable:
        layer.trainable = True
    else:
        layer.trainable = False

0x03 微调模型

#构建ImageDataGenerator生成训练和验证集
import os
base_dir = 'D:\\Jupyter\\dogs-vs-cats\\'
train_dir = os.path.join(base_dir, 'train')
validation_dir = os.path.join(base_dir, 'validation')
test_dir = os.path.join(base_dir, 'test')

from keras.preprocessing.image import ImageDataGenerator
from keras import optimizers

train_datagen = ImageDataGenerator(
    rescale = 1./255,
    rotation_range = 40,
    width_shift_range = 0.2,
    height_shift_range = 0.2,
    shear_range = 0.2,
    zoom_range = 0.2,
    horizontal_flip = True,
    fill_mode = 'nearest'
)

test_datagen = ImageDataGenerator(rescale = 1./255)

train_generator = train_datagen.flow_from_directory(
    train_dir,
    target_size = (150, 150),
    batch_size = 20,
    class_mode = 'binary',#因为使用binary_crossentropy，所以得使用binary二进制标签
)

validation_generator = test_datagen.flow_from_directory(
    validation_dir,
    target_size = (150,150),
    batch_size = 20,
    class_mode = 'binary',
)
#构建完整模型
from keras import models
from keras import layers

model = models.Sequential()
model.add(conv_base)
model.add(layers.Flatten())
model.add(layers.Dense(256, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))

model.compile(optimizer=optimizers.RMSprop(lr=2e-5),
             loss='binary_crossentropy',
             metrics=['acc'])
#训练
history = model.fit_generator(
    train_generator,
    steps_per_epoch=100,
    epochs=100,
    validation_data = validation_generator,
    validation_steps=50
)

0x04 评估

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 accuary')
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()

#最后可以保存模型
#model.save('cats_and_dogs_VGG16_unfreeze.h5')

完整参见