organize_image_folder.py

import logging

from featurepred.data import ImageFolderBuilder


def generate_folders_per_feature(positive_image_folder: str, negative_image_folder: str,
                                 training_split: bool, attribute_id: int):
    image_directory: str = 'data/birds/CUB_200_2011/images/'
    attributes_data_file: str = 'data/birds/CUB_200_2011/attributes/image_attribute_labels.txt'
    images_data_file: str = 'data/birds/CUB_200_2011/images.txt'
    split_data_file: str = 'data/birds/CUB_200_2011/train_test_split.txt'
    probably_certainty_id: int = 3
    image_folder_builder: ImageFolderBuilder = ImageFolderBuilder(attributes_data_file=attributes_data_file,
                                                                  images_data_file=images_data_file,
                                                                  image_directory=image_directory,
                                                                  split_data_file=split_data_file)
    image_folder_builder.build(attribute_id=attribute_id, minimum_certainty_id=probably_certainty_id,
                               positive_image_folder=positive_image_folder, negative_image_folder=negative_image_folder,
                               is_training=training_split)


if __name__ == "__main__":
    logging.basicConfig(level=logging.INFO)

    size_large_attribute_id: int = 218

    generate_folders_per_feature(positive_image_folder='data/feature_data/train/feature_positive',
                                 negative_image_folder='data/feature_data/train/feature_negative',
                                 training_split=True,
                                 attribute_id=size_large_attribute_id)

    generate_folders_per_feature(positive_image_folder='data/feature_data/val/feature_positive',
                                 negative_image_folder='data/feature_data/val/feature_negative',
                                 training_split=False,
                                 attribute_id=size_large_attribute_id)