Intro

这是深度学习第5课

在本课程结束时，您将能够使用数据增强。这个技巧让你看起来拥有的数据远远超过实际拥有的数据，从而产生更好的模型。

Lesson

[1]

from IPython.display import YouTubeVideo
YouTubeVideo('ypt_BAotCLo', width=800, height=450)

Sample Code

我们有一些你以前见过的模型设置代码。它暂时不是我们关注的焦点.

[2]

from tensorflow.python.keras.applications import ResNet50
from tensorflow.python.keras.models import Sequential
from tensorflow.python.keras.layers import Dense, Flatten, GlobalAveragePooling2D

num_classes = 2
resnet_weights_path = '../input/resnet50/resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'

my_new_model = Sequential()
my_new_model.add(ResNet50(include_top=False, pooling='avg', weights=resnet_weights_path))
my_new_model.add(Dense(num_classes, activation='softmax'))

# Say not to train first layer (ResNet) model. It is already trained
my_new_model.layers[0].trainable = False

my_new_model.compile(optimizer='sgd', loss='categorical_crossentropy', metrics=['accuracy'])

Fitting a Model With Data Augmentation

[3]

from tensorflow.python.keras.applications.resnet50 import preprocess_input
from tensorflow.python.keras.preprocessing.image import ImageDataGenerator

image_size = 224

data_generator_with_aug = ImageDataGenerator(preprocessing_function=preprocess_input,
                                   horizontal_flip=True,
                                   width_shift_range = 0.2,
                                   height_shift_range = 0.2)

train_generator = data_generator_with_aug.flow_from_directory(
        '../input/urban-and-rural-photos/rural_and_urban_photos/train',
        target_size=(image_size, image_size),
        batch_size=24,
        class_mode='categorical')

data_generator_no_aug = ImageDataGenerator(preprocessing_function=preprocess_input)
validation_generator = data_generator_no_aug.flow_from_directory(
        '../input/urban-and-rural-photos/rural_and_urban_photos/val',
        target_size=(image_size, image_size),
        class_mode='categorical')

my_new_model.fit_generator(
        train_generator,
        steps_per_epoch=3,
        epochs=2,
        validation_data=validation_generator,
        validation_steps=1)

Found 72 images belonging to 2 classes.
Found 20 images belonging to 2 classes.
Epoch 1/2
3/3 [==============================] - 32s 11s/step - loss: 0.7974 - acc: 0.5556 - val_loss: 0.9505 - val_acc: 0.7000
Epoch 2/2
3/3 [==============================] - 28s 9s/step - loss: 0.5379 - acc: 0.7639 - val_loss: 0.4675 - val_acc: 0.8000

<tensorflow.python.keras._impl.keras.callbacks.History at 0x7f8ce4375f60>

Exercise:Data Augmentation

Exercise Introduction

我们将返回您在上一个练习中处理的自动旋转问题。
我们还提供了您已经使用过的大部分代码。复制这篇笔记并采取数据增加步骤（下面的步骤2）。

1) Specify and Compile the Model

这与您之前使用的代码的工作方式相同。所以你在这里收到了它的完整版本。运行此单元格以指定和编译模型

【4】

from tensorflow.python.keras.applications import ResNet50
from tensorflow.python.keras.models import Sequential
from tensorflow.python.keras.layers import Dense, Flatten, GlobalAveragePooling2D

num_classes = 2
resnet_weights_path = '../input/resnet50/resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'

my_new_model = Sequential()
my_new_model.add(ResNet50(include_top=False, pooling='avg', weights=resnet_weights_path))
my_new_model.add(Dense(num_classes, activation='softmax'))

my_new_model.layers[0].trainable = False

my_new_model.compile(optimizer='sgd', loss='categorical_crossentropy', metrics=['accuracy'])

2) Fit the Model Using Data Augmentation

填写空白，并取消注释这些代码行。在这样做之后，您可以运行此单元格，您应该获得一个达到约90％准确度的模型。通过使用数据扩充，您可以将错误率降低一半。

【5】

from tensorflow.python.keras.applications.resnet50 import preprocess_input
from tensorflow.python.keras.preprocessing.image import ImageDataGenerator

image_size = 224

# Specify the values for all arguments to data_generator_with_aug. Then uncomment those lines
#data_generator_with_aug = ImageDataGenerator(preprocessing_function=preprocess_input
#                                              horizontal_flip = _____,
#                                              width_shift_range = ____,
#                                              height_shift_range = ____)
            
# data_generator_no_aug = ImageDataGenerator(preprocessing_function=preprocess_input)


# Specify which type of ImageDataGenerator above is to load in training data
#train_generator = ____.flow_from_directory(
#        directory = '../input/dogs-gone-sideways/images/train',
#        target_size=(image_size, image_size),
#        batch_size=12,
#        class_mode='categorical')

# Specify which type of ImageDataGenerator above is to load in validation data
#validation_generator = ____.flow_from_directory(
#        directory = '../input/dogs-gone-sideways/images/val',
#        target_size=(image_size, image_size),
#        class_mode='categorical')

#my_new_model.fit_generator(
#        ____, # specify where model gets training data
#        epochs = 3,
#        steps_per_epoch=19,
#        validation_data=____) # specify where model gets validation data

5, Data Augmentation