galleries 0.4.4

Package to work with galleries of images

Galleries

Galleries is a python package to manage images galleries and is mainly designed for recognition purposes. With this package you can create a gallery specifying where to get the images from and how to get annotations (if exist) from each image.

Installation

pip install galleries

Usage

Gallery construction

from galleries.annotations_parsers.file_name_parser import FileNameSepParser
from galleries.gallery import Gallery
from galleries.images_providers.local_files_image_providers import LocalFilesImageProvider


images_provider = LocalFilesImageProvider("path/to/images")
annotations_parser = FileNameSepParser(["label", "age"], sep="_")
gallery = Gallery(images_provider, annotations_parser)

Traverse images

images = gallery.get_images()  # returns a generator
for image in images:
  # image is a numpy ndarray
  pass

Get annotations

annotations = gallery.get_annotations()  # returns a generator
for annotation in annotations:
  # annotation is a dictionary
  pass

You can also generate new data for each image and easily save it using a GalleryDataHandler, which is an abstract class. GalleryGenericDataHandler is an implementation of this class that takes information of how to generate the new data and where to save it.

For instance, this is a code example of how to extract features from a gallery and save it to disk:

from galleries.write_gallery_data import GalleryGenericDataHandler

feature_extractor = ...  # your feature extractor here which has a features(image) method
data_generator = "<folder name>", "<feature extractor name>", feature_extractor.features

# write features if do not exist
gallery_features_writer = GalleryGenericDataHandler(gallery, "directory/to/save/features")
exist_features = gallery_features_writer.exists_data(data_generator)
if not exist_features:
  gallery_features_writer.write_data(data_generator)

# read features
features = gallery_features_writer.read_data(data_generator)