如何在Keras中格式化ImageDataGenerator中的X和Y数据？

Question

现在，我能够在图像数据上训练和测试神经网络。我希望能够从DirectoryIterator train_generator中提取图像数据和相应的标签，以便也许可以改组此数据并执行交叉验证。有没有一种方法可以提取（如果需要的话可以格式化）这些数据？

from keras.preprocessing.image import ImageDataGenerator

img_width, img_height = 150, 150
train_data_dir = '/train'
datagen = ImageDataGenerator(rescale=1./255) #rescales [0,1]
train_generator = datagen.flow_from_directory(train_data_dir, target_size=(img_width, img_height), shuffle=True, batch_size=32, class_mode='binary')
#I want to do something like (X, Y) = train_generator.getData()

Answer 1

train_generator = datagen.flow_from_directory(
   train_data_dir, 
   target_size=(img_width, img_height), 
   shuffle=True, 
   batch_size=SIZE_OF_DATASET,  ## change to size of entire dataset
   class_mode='binary')

for X, Y in train_generator:
   break