BSD100.py

# coding=utf-8
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""BSD100 dataset: An evaluation dataset for the image super resolution task"""


import datasets
from pathlib import Path


_CITATION = """
@inproceedings{martin2001database,
  title={A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics},
  author={Martin, David and Fowlkes, Charless and Tal, Doron and Malik, Jitendra},
  booktitle={Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001},
  volume={2},
  pages={416--423},
  year={2001},
  organization={IEEE}
}
"""

_DESCRIPTION = """
BSD is a dataset used frequently for image denoising and super-resolution. 
BSD100 is the testing set of the Berkeley segmentation dataset BSD300.
"""

_HOMEPAGE = "https://www2.eecs.berkeley.edu/Research/Projects/CS/vision/bsds/"

_LICENSE = "UNK"

_DL_URL = "https://huggingface.co/datasets/eugenesiow/BSD100/resolve/main/data/"

_DEFAULT_CONFIG = "bicubic_x2"

_DATA_OPTIONS = {
    "bicubic_x2": {
        "hr": _DL_URL + "BSD100_HR.tar.gz",
        "lr": _DL_URL + "BSD100_LR_x2.tar.gz",
    },
    "bicubic_x3": {
        "hr": _DL_URL + "BSD100_HR.tar.gz",
        "lr": _DL_URL + "BSD100_LR_x3.tar.gz",
    },
    "bicubic_x4": {
        "hr": _DL_URL + "BSD100_HR.tar.gz",
        "lr": _DL_URL + "BSD100_LR_x4.tar.gz",
    }
}


class Bsd100Config(datasets.BuilderConfig):
    """BuilderConfig for BSD100."""

    def __init__(
        self,
        name,
        hr_url,
        lr_url,
        **kwargs,
    ):
        if name not in _DATA_OPTIONS:
            raise ValueError("data must be one of %s" % _DATA_OPTIONS)
        super(Bsd100Config, self).__init__(name=name, version=datasets.Version("1.0.0"), **kwargs)
        self.hr_url = hr_url
        self.lr_url = lr_url


class Bsd100(datasets.GeneratorBasedBuilder):
    """BSD100 dataset for single image super resolution evaluation."""

    BUILDER_CONFIGS = [
        Bsd100Config(
            name=key,
            hr_url=values['hr'],
            lr_url=values['lr']
        ) for key, values in _DATA_OPTIONS.items()
    ]

    DEFAULT_CONFIG_NAME = _DEFAULT_CONFIG

    def _info(self):
        features = datasets.Features(
            {
                "hr": datasets.Value("string"),
                "lr": datasets.Value("string"),
            }
        )
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            supervised_keys=None,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        hr_data_dir = dl_manager.download_and_extract(self.config.hr_url)
        lr_data_dir = dl_manager.download_and_extract(self.config.lr_url)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "lr_path": lr_data_dir,
                    "hr_path": str(Path(hr_data_dir) / 'BSD100_HR')
                },
            )
        ]

    def _generate_examples(
        self, hr_path, lr_path
    ):
        """ Yields examples as (key, example) tuples. """
        # This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
        # The `key` is here for legacy reason (tfds) and is not important in itself.
        extensions = {'.png'}
        for file_path in sorted(Path(lr_path).glob("**/*")):
            if file_path.suffix in extensions:
                file_path_str = str(file_path.as_posix())
                yield file_path_str, {
                    'lr': file_path_str,
                    'hr': str((Path(hr_path) / file_path.name).as_posix())
                }