SynWBM.py

# 
 # This file is part of the SynWBM distribution (https://huggingface.co/datasets/ABC-iRobotics/SynWBM).
 # Copyright (c) 2023 ABC-iRobotics.
 # 
 # This program is free software: you can redistribute it and/or modify  
 # it under the terms of the GNU General Public License as published by  
 # the Free Software Foundation, version 3.
 #
 # This program is distributed in the hope that it will be useful, but 
 # WITHOUT ANY WARRANTY; without even the implied warranty of 
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 
 # General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License 
 # along with this program. If not, see <http://www.gnu.org/licenses/>.
 #
"""SynWBM dataset"""

import sys
if sys.version_info < (3, 9):
    from typing import Sequence, Generator, Tuple
else:
    from collections.abc import Sequence, Generator
    Tuple = tuple

from typing import Optional, IO

import datasets
import itertools


# ---- Constants ----

_CITATION = """\
COMING SOON
"""

_DESCRIPTION = """\
A synthetic instance segmentation dataset for white button mushrooms (Agaricus bisporus).
The dataset incorporates rendered and generated synthetic images for training mushroom segmentation models.
"""

_HOMEPAGE = "https://huggingface.co/datasets/ABC-iRobotics/SynWBM"

_LICENSE = "GNU General Public License v3.0"

_LATEST_VERSIONS = {
    "all": "1.0.0",
    "blender": "1.0.0",
    "sdxl": "1.0.0",
}

BASE_URL = "https://huggingface.co/datasets/ABC-iRobotics/SynWBM/resolve/main/"



# ---- SynWBM dataset Configs ----

class SynWBMDatasetConfig(datasets.BuilderConfig):
    """BuilderConfig for SynWBM dataset."""

    def __init__(self, name: str, base_urls: Sequence[str], images_txt: str, version: Optional[str] = None, **kwargs):
        _version = _LATEST_VERSIONS[name] if version is None else version
        super(SynWBMDatasetConfig, self).__init__(version=datasets.Version(_version), name=name, **kwargs)
        with open(images_txt, 'r') as f:
            image_list = f.readlines()
        img_urls = []
        depth_urls = []
        mask_urls = []
        for base_url in base_urls:
            img_urls.extend([base_url + image.strip() for image in image_list])
            depth_urls.extend([BASE_URL + "depths/" + image.strip() for image in image_list])
            mask_urls.extend([BASE_URL + "masks/" + image.strip() for image in image_list])

        self._imgs_urls = img_urls
        self._depth_urls = depth_urls
        self._masks_urls = mask_urls


    @property
    def features(self):
        return datasets.Features(
            {
                "image": datasets.Image(),
                "depth": datasets.Image(),
                "mask": datasets.Image(),
            }
        )
    
    @property
    def supervised_keys(self):
        return None



# ---- SynWBM dataset Loader ----

class SynWBMDataset(datasets.GeneratorBasedBuilder):
    """SynWBM dataset."""

    BUILDER_CONFIG_CLASS = SynWBMDatasetConfig
    BUILDER_CONFIGS = [
        SynWBMDatasetConfig(
            name = "all",
            description = "All images",
            base_urls = [
                BASE_URL + "rendered/",
                BASE_URL + "generated/"
            ],
            images_txt = "images.txt"
        ),
        SynWBMDatasetConfig(
            name = "blender",
            description = "Synthetic images rendered using Blender",
            base_urls = [
                BASE_URL + "rendered/"
            ],
            images_txt = "images.txt"
        ),
        SynWBMDatasetConfig(
            name = "sdxl",
            description = "Synthetic images generated by Stable Diffusion XL",
            base_urls = [
                BASE_URL + "generated/"
            ],
            images_txt = "images.txt"
        ),
    ]
    DEFAULT_WRITER_BATCH_SIZE = 10

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=self.config.features,
            supervised_keys=self.config.supervised_keys,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
            version=self.config.version,
        )

    def _split_generators(self, dl_manager):
        imgs_paths = dl_manager.download(self.config._imgs_urls)
        depths_paths = dl_manager.download(self.config._depth_urls)
        masks_paths = dl_manager.download(self.config._masks_urls)

        imgs_gen = itertools.chain.from_iterable([dl_manager.iter_archive(path) for path in imgs_paths])
        depths_gen = itertools.chain.from_iterable([dl_manager.iter_archive(path) for path in depths_paths])
        masks_gen = itertools.chain.from_iterable([dl_manager.iter_archive(path) for path in masks_paths])
        
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "images": imgs_gen,
                    "depths": depths_gen,
                    "masks": masks_gen,
                },
            ),
        ]

    def _generate_examples(
        self,
        images: Generator[Tuple[str,IO], None, None],
        depths: Generator[Tuple[str,IO], None, None],
        masks: Generator[Tuple[str,IO], None, None],
    ):
        for i, (img_info, depth_info, mask_info) in enumerate(zip(images, depths, masks)):
            img_file_path, img_file_obj = img_info
            depth_file_path, depth_file_obj = depth_info
            mask_file_path, mask_file_obj = mask_info

            img_bytes = img_file_obj.read()
            depth_bytes = depth_file_obj.read()
            mask_bytes = mask_file_obj.read()
            img_file_obj.close()
            depth_file_obj.close()
            mask_file_obj.close()

            yield i, {
                "image": {"path": img_file_path, "bytes": img_bytes},
                "depth": {"path": depth_file_path, "bytes": depth_bytes},
                "mask": {"path": mask_file_path, "bytes": mask_bytes},
            }