def __init__(self, root_folder, metadata_file="labels.json",
description="Generic Dataset JSON-Based of image with Age labels", dataset_normalizers=None):
"""
Initialization of a dataset of image with ages labeled.
Metadata is built from a JSON file.
:param root_folder:
:param metadata_file: JSON file URI, which is composed by the following format: {'IMAGE_FILE_NAME': AGE_RANGE_STRING}
Example:
{'test/image1.jpg': '[23,34]'}
Note: the working directory when loading the metadata_file is root_folder.
:param description: description of the dataset for report purposes.
:param dataset_normalizers: list of normalizers to normalize when storing image inside this dataset.
:return:
"""
# This dataset class is also capable of creating datasets.
mkdir_p(root_folder)
# When metadata_file is not an absolute URI, it is related to root_folder.
if not self._is_absolute_uri(metadata_file):
metadata_file = os.path.join(root_folder, metadata_file)
Dataset.__init__(self, root_folder, metadata_file, description)
self.autoencoded_uris = {}
if not dataset_normalizers:
dataset_normalizers = []
self.normalizers = dataset_normalizers
def put_image(self, image, autoencode_uri=True, apply_normalizers=True):
"""
Puts an image in the dataset.
It must be filled with content, relative uri and metadata in order to be created the dataset.
:param image: Image to save in the dataset.
:param autoencode_uri: Boolean flag to set if the URI should be automatically filled by the dataset or not.
:param dataset_normalizer: normalizer to apply to the image
:param apply_normalizers: boolean flag to apply normalizers when the image is put into the dataset manually.
:return:
"""
if autoencode_uri:
uri = self._encode_uri_for_image(image)
else:
uri = image.get_uri()
if self._is_absolute_uri(uri):
raise Exception(
"Uri for storing into dataset must be relative, not absolute")
key = uri # We index by the relative uri
uri = os.path.join(self.root_folder, uri)
mkdir_p(os.path.dirname(uri))
self.metadata_content[key] = image.get_metadata()[0]
try:
if not image.is_loaded():
image.load_from_uri()
if not image.is_loaded():
raise Exception("Image may not exist or it is not valid.")
image_blob = image.get_blob()
normalizers_applied = 0
if apply_normalizers:
for normalizer in self.normalizers:
image_blob = normalizer.apply(image_blob)
normalizers_applied += 1
cv2.imwrite(uri, image_blob)
print("Saved into {} ({} normalizers applied)".format(
uri, normalizers_applied))
except Exception as ex:
print(
"Could not write image \"{}\" into dataset.Reason: {}".format(
image.get_uri(), ex))
del self.metadata_content[key]