如何在 Keras 中正确使用 ImageDataGenerator?

How to correctly use ImageDataGenerator in Keras?

我最近在 Keras 中玩数据扩充,我正在使用基本的 ImageDataGenerator。我通过艰难的方式了解到它实际上是一个生成器,而不是迭代器(因为 type(train_aug_ds) 给出 <class 'keras.preprocessing.image.DirectoryIterator'> 我认为它是一个迭代器)。我也查看了一些关于使用它的博客,但他们没有回答我所有的问题。

所以,我这样加载了我的数据:

train_aug = ImageDataGenerator(
    rescale=1./255,
    horizontal_flip=True,
    height_shift_range=0.1,
    width_shift_range=0.1,
    brightness_range=(0.5,1.5),
    zoom_range = [1, 1.5],
)
train_aug_ds = train_aug.flow_from_directory(
    directory='./train',
    target_size=image_size,
    batch_size=batch_size,
)

为了训练我的模型,我执行了以下操作:

model.fit(
    train_aug_ds,
    epochs=150,
    validation_data=(valid_aug_ds,),
)

它奏效了。我有点困惑它是如何工作的,因为 train_aug_ds 是生成器,所以它应该提供无限大的数据集。文档说:

When passing an infinitely repeating dataset, you must specify the steps_per_epoch argument.

我没做过,但它起作用了。它是否以某种方式推断出步数?另外,它是只使用增强数据,还是批量使用非增强图像?

所以基本上,我的问题是如何正确使用这个生成器和函数 fit 来获得我训练集中的所有数据,包括原始的、非增强的图像和增强的图像,并循环使用它几个times/steps(目前看来每个 epoch 只执行一步)?

我认为文档可能非常混乱,而且我认为行为会因您的 Tensorflow 和 Keras 版本而异。例如,在这个 , the user is describing the exact behavior you are expecting. Generally, the flow_from_directory() method allows you to read the images directly from a directory and augment them while your model is being trained and as already stated 中,它在每个时期对每个文件夹中的每个样本进行迭代。使用以下示例,您可以通过查看进度条中每个时期的步数来检查是否属于这种情况(在 TF 2.7 上):

import tensorflow as tf

BATCH_SIZE = 64

flowers = tf.keras.utils.get_file(
    'flower_photos',
    'https://storage.googleapis.com/download.tensorflow.org/example_images/flower_photos.tgz',
    untar=True)

img_gen = tf.keras.preprocessing.image.ImageDataGenerator(
    rescale=1./255,
    horizontal_flip=True,
)

train_ds = img_gen.flow_from_directory(flowers, batch_size=BATCH_SIZE, shuffle=True, class_mode='sparse')
num_classes = 5

model = tf.keras.Sequential([
  tf.keras.layers.Conv2D(16, 3, padding='same', activation='relu', input_shape=(256, 256, 3)),
  tf.keras.layers.MaxPooling2D(),
  tf.keras.layers.Conv2D(32, 3, padding='same', activation='relu'),
  tf.keras.layers.MaxPooling2D(),
  tf.keras.layers.Conv2D(64, 3, padding='same', activation='relu'),
  tf.keras.layers.MaxPooling2D(),
  tf.keras.layers.Flatten(),
  tf.keras.layers.Dense(128, activation='relu'),
  tf.keras.layers.Dense(num_classes)
])

model.compile(optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True))

epochs=10
history = model.fit(
  train_ds,
  epochs=epochs
)
Found 3670 images belonging to 5 classes.
Epoch 1/10
 6/58 [==>...........................] - ETA: 3:02 - loss: 2.0608

如果像这样用 tf.data.Dataset.from_generator 包裹 flow_from_directory

train_ds = tf.data.Dataset.from_generator(
    lambda: img_gen.flow_from_directory(flowers, batch_size=BATCH_SIZE, shuffle=True, class_mode='sparse'),
    output_types=(tf.float32, tf.float32))

您会注意到进度条看起来像这样,因为 steps_per_epoch 尚未明确定义:

Epoch 1/10
Found 3670 images belonging to 5 classes.
     29/Unknown - 104s 4s/step - loss: 2.0364

如果你添加这个参数,你会在进度条中看到步骤:

history = model.fit(
  train_ds,
  steps_per_epoch = len(from_directory),
  epochs=epochs
)
Found 3670 images belonging to 5 classes.
Epoch 1/10
 3/58 [>.............................] - ETA: 3:19 - loss: 4.1357

最后,对于你的问题:

How to use this generator correctly with function fit to have all data in my training set, including original, non-augmented images and augmented images, and to cycle through it several times/step?

您可以简单地将 steps_per_epoch 增加到 number of samples // batch_size 以上,方法是乘以某个因数:

history = model.fit(
  train_ds,
  steps_per_epoch = len(from_directory)*2,
  epochs=epochs
)
Found 3670 images belonging to 5 classes.
Epoch 1/10
  1/116 [..............................] - ETA: 12:11 - loss: 1.5885

现在每个纪元有 116 步,而不是 58 步。