Error resulting from ImageDataGenerator during data augmentation - tensorflow

Question

Can someone please help me in fixing the error? The code works fine before the for loop. Before the for loop, an array of the image was printed. Is there something wrong with the for loop? The output should be a file stored with augmented images of the input image. The input image is a jpg image.
The code I wrote:
import keras
import tensorflow as tf
from keras.preprocessing.image import ImageDataGenerator
data_gen = tf.keras.preprocessing.image.ImageDataGenerator(
rotation_range=45,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='contrast',
cval=125
)
x = io.imread('mona.jpg')
x = x.reshape((1, ) + x.shape) #Array with shape (1, 256, 256, 3)
i = 0
for batch in data_gen.flow(x, batch_size=16, save_to_dir='/Users/ghad/Desktop',
save_prefix='aug',
save_format='jpg'):
i += 1
if i > 20:
The generated error:
RuntimeError Traceback (most recent call last)
Input In [14], in <cell line: 31>()
28 x = x.reshape((1, ) + x.shape) #Array with shape (1, 256, 256, 3)
30 i = 0
---> 31 for batch in data_gen.flow(x, batch_size=16,
32 save_to_dir='/Users/ghadahalhabib/Desktop',
33 save_prefix='aug',
34 save_format='jpg'):
35 i += 1
36 if i > 20:
File ~/opt/anaconda3/envs/tensorflow/lib/python3.9/site-packages/keras/preprocessing/image.py:148, in Iterator.__next__(self, *args, **kwargs)
147 def __next__(self, *args, **kwargs):
--> 148 return self.next(*args, **kwargs)
File ~/opt/anaconda3/envs/tensorflow/lib/python3.9/site-packages/keras/preprocessing/image.py:160, in Iterator.next(self)
157 index_array = next(self.index_generator)
158 # The transformation of images is not under thread lock
159 # so it can be done in parallel
--> 160 return self._get_batches_of_transformed_samples(index_array)
File ~/opt/anaconda3/envs/tensorflow/lib/python3.9/site-packages/keras/preprocessing/image.py:709, in NumpyArrayIterator._get_batches_of_transformed_samples(self, index_array)
707 x = self.x[j]
708 params = self.image_data_generator.get_random_transform(x.shape)
--> 709 x = self.image_data_generator.apply_transform(
710 x.astype(self.dtype), params)
711 x = self.image_data_generator.standardize(x)
712 batch_x[i] = x
File ~/opt/anaconda3/envs/tensorflow/lib/python3.9/site-packages/keras/preprocessing/image.py:1800, in ImageDataGenerator.apply_transform(self, x, transform_parameters)
1797 img_col_axis = self.col_axis - 1
1798 img_channel_axis = self.channel_axis - 1
-> 1800 x = apply_affine_transform(
1801 x,
1802 transform_parameters.get('theta', 0),
1803 transform_parameters.get('tx', 0),
1804 transform_parameters.get('ty', 0),
1805 transform_parameters.get('shear', 0),
1806 transform_parameters.get('zx', 1),
1807 transform_parameters.get('zy', 1),
1808 row_axis=img_row_axis,
1809 col_axis=img_col_axis,
1810 channel_axis=img_channel_axis,
1811 fill_mode=self.fill_mode,
1812 cval=self.cval,
1813 order=self.interpolation_order)
1815 if transform_parameters.get('channel_shift_intensity') is not None:
1816 x = apply_channel_shift(x,
1817 transform_parameters['channel_shift_intensity'],
1818 img_channel_axis)
File ~/opt/anaconda3/envs/tensorflow/lib/python3.9/site-packages/keras/preprocessing/image.py:2324, in apply_affine_transform(x, theta, tx, ty, shear, zx, zy, row_axis, col_axis, channel_axis, fill_mode, cval, order)
2321 final_affine_matrix = transform_matrix[:2, :2]
2322 final_offset = transform_matrix[:2, 2]
-> 2324 channel_images = [ndimage.interpolation.affine_transform( # pylint: disable=g-complex-comprehension
2325 x_channel,
2326 final_affine_matrix,
2327 final_offset,
2328 order=order,
2329 mode=fill_mode,
2330 cval=cval) for x_channel in x]
2331 x = np.stack(channel_images, axis=0)
2332 x = np.rollaxis(x, 0, channel_axis + 1)
File ~/opt/anaconda3/envs/tensorflow/lib/python3.9/site-packages/keras/preprocessing/image.py:2324, in <listcomp>(.0)
2321 final_affine_matrix = transform_matrix[:2, :2]
2322 final_offset = transform_matrix[:2, 2]
-> 2324 channel_images = [ndimage.interpolation.affine_transform( # pylint: disable=g-complex-comprehension
2325 x_channel,
2326 final_affine_matrix,
2327 final_offset,
2328 order=order,
2329 mode=fill_mode,
2330 cval=cval) for x_channel in x]
2331 x = np.stack(channel_images, axis=0)
2332 x = np.rollaxis(x, 0, channel_axis + 1)
File ~/opt/anaconda3/envs/tensorflow/lib/python3.9/site-packages/scipy/ndimage/interpolation.py:574, in affine_transform(input, matrix, offset, output_shape, output, order, mode, cval, prefilter)
572 npad = 0
573 filtered = input
--> 574 mode = _ni_support._extend_mode_to_code(mode)
575 matrix = numpy.asarray(matrix, dtype=numpy.float64)
576 if matrix.ndim not in [1, 2] or matrix.shape[0] < 1:
File ~/opt/anaconda3/envs/tensorflow/lib/python3.9/site-packages/scipy/ndimage/_ni_support.py:54, in _extend_mode_to_code(mode)
52 return 6
53 else:
---> 54 raise RuntimeError('boundary mode not supported')
RuntimeError: boundary mode not supported

Answer 1

for the code
for batch in data_gen.flow(x, batch_size=16, save_to_dir='/Users/ghad/Desktop', save_prefix='aug', save_format='jpg'):
you are inputting only a single image but asking to produce 16 augmented images. That won't work. Normal the length of x is LARGER than the batch size. Set the batch size to 1. That way you will produce 1 augment image each time you feed a new image into the generator