How to Solve: 'str' object has no attribute 'data_format' in keras

Question

I am trying to make a classifier which can classify cats and dogs using keras. I am just trying to create the tensor data from images using ImageDataGenerator.flow_from_directory() which are sorted and kept in the directories whose paths are given in train_path, test_path etc.

Here is my code:

import numpy as np

import keras

from keras import backend as K

from keras.models import Sequential

from keras.layers import Activation



train_path = 'cats-and-dogs/train' test_path = 'cats-and-dogs/test' valid_path = 'cats-and-dogs/valid'

train_dir = 'cats-and-dogs/' test_dir = 'cats-and-dogs/' valid_dir = 'cats-and-dogs/'



train_batches = ImageDataGenerator.flow_from_directory(train_path, directory=train_dir, target_size=(200,200), classes=['dog','cat'], batch_size=10)

test_batches = ImageDataGenerator.flow_from_directory(test_path, directory=test_dir, target_size=(200,200), classes=['dog','cat'], batch_size=5)

valid_batches = ImageDataGenerator.flow_from_directory(valid_path, directory=valid_dir, target_size=(200,200), classes=['dog','cat'], batch_size=10)

But I am getting the following error using python 3.5:

/usr/local/lib/python3.5/site-packages/h5py/init.py:36: FutureWarning: Conversion of the second argument of issubdtype from float to np.floating is deprecated. In future, it will be treated as np.float64 == np.dtype(float).type. from ._conv import register_converters as _register_converters Using TensorFlow backend. Traceback (most recent call last): File "CNNFromScratch.py", line 29, in train_batches = ImageDataGenerator.flow_from_directory(train_path, directory=train_dir, target_size=(200,200), classes=['dog','cat'], batch_size=10) File "/usr/local/lib/python3.5/site-packages/keras/preprocessing/image.py", line 565, in flow_from_directory data_format=self.data_format,

AttributeError: 'str' object has no attribute 'data_format'

What can I do to solve this problem?

Answer 1

Method flow_from_directory of ImageDataGenerator is not static. Therefore you first have to initialize an instance of class ImageDataGenerator and then call this method.

This should work:

import numpy as np

import keras

from keras import backend as K

from keras.models import Sequential
from keras.layers import Activation
from keras.preprocessing.image import ImageDataGenerator

train_path = 'cats-and-dogs/train'
test_path = 'cats-and-dogs/test'
valid_path = 'cats-and-dogs/valid'

my_generator = ImageDataGenerator()

train_batches = my_generator.flow_from_directory(directory=train_path, target_size=(200,200), classes=['dog','cat'], batch_size=10)

test_batches = my_generator.flow_from_directory(directory=test_path, target_size=(200,200), classes=['dog','cat'], batch_size=5)

valid_batches = my_generator.flow_from_directory(directory=valid_path, target_size=(200,200), classes=['dog','cat'], batch_size=10)

Check documentation for adding more parameters.