images.py

import os

import datasets
from datasets.tasks import ImageClassification


_URLS = {
    "train": "https://huggingface.co/datasets/bansilp/bansilp/mcl_r/blob/main/train.zip",
    
    }

_NAMES = [
    "blues",
    "classical",
    "country",
    "disco",
    "hiphop",
    "metal",
    "pop",
    "reggae",
    "rock"
]


class uta_rldd(datasets.GeneratorBasedBuilder):
   

    def _info(self):
        return datasets.DatasetInfo(
            features=datasets.Features(
                {
                    "image": datasets.Image(),
                    "labels": datasets.features.ClassLabel(names=_NAMES),
                }
            ),
            supervised_keys=("image", "labels"),
            task_templates=[ImageClassification(image_column="image", label_column="labels")],
        )

    def _split_generators(self, dl_manager):
        data_files = dl_manager.download_and_extract(_URLS)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "files": dl_manager.iter_files([data_files["train"]]),
                },
            )
        ]

    def _generate_examples(self, files):
        for i, path in enumerate(files):
            file_name = os.path.basename(path)
            if file_name.endswith(".png"):
                yield i, {
                    "image": path,
                    "labels": os.path.basename(os.path.dirname(path)).lower(),
                }