import os
from keras.preprocessing.image import ImageDataGenerator, array_to_img, img_to_array, load_img
from PIL import Image
import glob



image_count = 5
current_dataset_folder = "train/*.*"
export_folder_name = "newdataset"
export_image_format = "jpeg"

keras_data_generator = ImageDataGenerator(rotation_range=40,
                             width_shift_range=0.2,
                             height_shift_range=0.2,
                             shear_range=0.2,
                             zoom_range=0.2,
                             horizontal_flip=True,
                             vertical_flip=True,
                             fill_mode='nearest')


for filename in glob.glob(current_dataset_folder):

    img = load_img(filename)
    x = img_to_array(img)
    x = x.reshape((1,) + x.shape)

    i = 0
    for batch in keras_data_generator.flow(x, batch_size=1, save_to_dir= export_folder_name, save_format= export_image_format):
      i += 1
      if i  > image_count :
        break