def test_get_keras_submodule_errors(monkeypatch):
with pytest.raises(ImportError):
keras_preprocessing.get_keras_submodule('something')
monkeypatch.setattr(keras_preprocessing, '_KERAS_BACKEND', None)
with pytest.raises(ImportError):
keras_preprocessing.get_keras_submodule('backend')
with pytest.raises(ImportError):
keras_preprocessing.get_keras_submodule('utils')
"""Utilities for real-time data augmentation on image data.
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import threading
import numpy as np
from keras_preprocessing import get_keras_submodule
try:
IteratorType = get_keras_submodule('utils').Sequence
except ImportError:
IteratorType = object
from .utils import (array_to_img, img_to_array, load_img)
from .affine_transformations import random_crop, center_crop, rotate_random_zoom_crop
class Iterator(IteratorType):
"""Base class for image data iterators.
Every `Iterator` must implement the `_get_batches_of_transformed_samples`
method.
# Arguments
n: Integer, total number of samples in the dataset to loop over.
batch_size: Integer, size of a batch.
shuffle: Boolean, whether to shuffle the data between epochs.
seed: Random seeding for data shuffling.
"""
import sys
import multiprocessing.pool
import numpy as np
import keras
import keras.preprocessing.image
from keras_preprocessing.image import Iterator, load_img, img_to_array
from keras_preprocessing import get_keras_submodule
backend = get_keras_submodule('backend')
keras_utils = get_keras_submodule('utils')
try:
from PIL import ImageEnhance
from PIL import Image as pil_image
except ImportError:
pil_image = None
ImageEnhance = None
try:
import scipy
except ImportError:
scipy = None
if pil_image is not None:
_PIL_INTERPOLATION_METHODS = {
'nearest': pil_image.NEAREST,
'bilinear': pil_image.BILINEAR,
'bicubic': pil_image.BICUBIC,
}
# These methods were only introduced in version 3.4.0 (2016).
if hasattr(pil_image, 'HAMMING'):
import os
import numpy as np
import PIL
from keras_preprocessing.image import ImageDataGenerator, DirectoryIterator, array_to_img, load_img, img_to_array
from keras_preprocessing import get_keras_submodule
backend = get_keras_submodule('backend')
class AugmentedDirectoryIterator(DirectoryIterator):
'''
AugmentedDirectoryIterator inherits from DirectoryIterator:
(https://github.com/keras-team/keras-preprocessing/blob/master/keras_preprocessing/image.py#L1811)
This implementation adds the functionality of computing multiple crops following the work Going Deeper with
Convolutions (https://arxiv.org/pdf/1409.4842.pdf) and allowing the use of transforms on such crops.
It includes the addition of data_augmentation as an argument. It is a dictionary consisting of 3 elements:
- 'scale_sizes': 'default' (4 similar scales to Original paper) or a list of sizes. Each scaled image then
will be cropped into three square parts. For each square, we then take the 4 corners and the center "target_size"
crop as well as the square resized to "target_size".
- 'transforms': list of transforms to apply to these crops in addition to not
applying any transform ('horizontal_flip', 'vertical_flip', 'rotate_90', 'rotate_180', 'rotate_270' are
supported now).
- 'crop_original': 'center_crop' mode allows to center crop the original image prior do the rest of transforms,
scalings + croppings.
If 'scale_sizes' is None the image will be resized to "target_size" and transforms will be applied over that image.
def test_get_keras_submodule(monkeypatch):
monkeypatch.setattr(keras_preprocessing, '_KERAS_BACKEND', 'backend')
assert 'backend' == keras_preprocessing.get_keras_submodule('backend')
monkeypatch.setattr(keras_preprocessing, '_KERAS_UTILS', 'utils')
assert 'utils' == keras_preprocessing.get_keras_submodule('utils')
"""Utilities for real-time data augmentation on image data.
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import threading
import numpy as np
from keras_preprocessing import get_keras_submodule
try:
IteratorType = get_keras_submodule('utils').Sequence
except ImportError:
IteratorType = object
from .utils import (array_to_img,
img_to_array,
load_img)
class Iterator(IteratorType):
"""Base class for image data iterators.
Every `Iterator` must implement the `_get_batches_of_transformed_samples`
method.
# Arguments
n: Integer, total number of samples in the dataset to loop over.
batch_size: Integer, size of a batch.
shuffle: Boolean, whether to shuffle the data between epochs.
seed: Random seeding for data shuffling.
