def test_multiple_context():
assert contexts.get_image_dir() == ''
assert contexts.get_image_col() == None
with ivs.plotting_context(image_col='test_col', image_dir='test_dir'):
assert contexts.get_image_dir() == 'test_dir'
assert contexts.get_image_col() == 'test_col'
assert contexts.get_image_dir() == ''
assert contexts.get_image_col() == None
def test_image_dir():
assert contexts.get_image_dir() == ''
assert contexts.get_image_col() == None
with ivs.plotting_context(image_dir='test'):
assert contexts.get_image_dir() == 'test'
assert contexts.get_image_col() == None
assert contexts.get_image_dir() == ''
assert contexts.get_image_col() == None
def directory_to_dataframe(image_dir='', features=None, n_jobs=-1):
"""Create a pandas.DataFrame containing the path to all images in
a directory.
The pandas.DataFrame has a single column `image_path`, which contains
the paths to the various images. The paths are relative to the given
`image_dir`. For example, a directory named `image_dir` with two
images `image1.jpg` and `image2.jpg` would result in a single
column dataframe. The column is named 'image_path' and it contains
the paths to the two images.
Parameters
----------
image_dir : str
The directory to search of images and place their paths
in a dataframe.
features : list or None
A list of features to include in the dataframe. The default (None)
includes no additional features.
n_jobs : int
The number of parallel jobs used to load the
images from disk.
Returns
-------
pandas.DataFrame
A dataframe containing the image paths.
"""
if not image_dir:
image_dir = contexts.get_image_dir()
image_files = list(
itertools.chain.from_iterable(
[image_glob(image_dir, ext) for ext in image_extensions]))
image_files = [f.split(image_dir + os.path.sep)[1] for f in image_files]
data = pd.DataFrame({'image_path': image_files})
if features:
if set(features) & set(feature_lib.HSVFeatures.all_features()):
images = load_images(data['image_path'],
image_dir=image_dir,
as_image=True,
n_jobs=n_jobs)
hsv = feature_lib.extract_hsv_stats(images, n_jobs=n_jobs)
for feature in features:
feature_idx = feature_lib.HSVFeatures.feature_index(feature)
data[feature] = hsv[:, feature_idx]
else:
raise ValueError('Unknown features.')
return data
def get_images(data, images,
image_dir='',
image_size=None,
as_image=False,
index=None,
n_jobs=1):
"""Helper function to load images from disk or properly format
an already existing image array.
Parameters
----------
data : pandas.DataFrame
Tidy ("long-form") dataframe where each column is a variable
and each row is an observation.
images : str or array-like of shape [n_samples, width, height, channels], optional
Image array or name of the variable containing the image file
paths within `data`.
image_dir : str, optional
The location of the image files on disk. Images will
be loaded from files matching the pattern
'image_dir + os.path.sep + image_path'.
image_size : int, optional
The size of each image displayed in the scatter plot. Images
will be sampled to `image_size` if the size of the images
do not match `image_size`.
as_image : bool
If True, the returned images are converted to PIL.Image
objects.
Returns
-------
images : array-like
Either a list of PIL.Images or a np.array of shape
[n_samples, width, height, channels].
"""
if isinstance(images, np.ndarray):
if as_image:
return [image_io.to_pillow_image(img, image_size=image_size) for
img in images]
elif image_size:
return np.asarray([scipy.misc.imresize(img, image_size, interp='lanczos') for
img in images])
return images
if not image_dir:
image_dir = contexts.get_image_dir()
if not images:
images = contexts.get_image_data()
if images is None:
images = data[contexts.get_image_col()]
if index is not None:
images = images.iloc[index]
images = image_io.load_images(
images,
image_dir=image_dir,
as_image=as_image,
image_size=image_size,
n_jobs=n_jobs)
return images