mhr_recognize_datasets.py

import os
import csv
import json
import datasets
import numpy as np
import pandas as pd
from pathlib import Path

_DESCRIPTION = """\
Monster Hunter Rise images and labels.
"""

_DATA_URL = {
    "whole_image": "https://huggingface.co/datasets/miojizzy/mhr_recognize_datasets/resolve/main/data/whole/train.zip",
    "whole_label": "https://huggingface.co/datasets/miojizzy/mhr_recognize_datasets/resolve/main/data/whole/label.csv",
    "hole_image": "https://huggingface.co/datasets/miojizzy/mhr_recognize_datasets/resolve/main/data/hole/train.zip",
    "hole_label": "https://huggingface.co/datasets/miojizzy/mhr_recognize_datasets/resolve/main/data/hole/label.csv",
    "skill_image": "https://huggingface.co/datasets/miojizzy/mhr_recognize_datasets/resolve/main/data/skill/train.zip",
    "skill_label": "https://huggingface.co/datasets/miojizzy/mhr_recognize_datasets/resolve/main/data/skill/label.csv",
}

class MHRRecognizeDatasetsConfig(datasets.BuilderConfig):
    def __init__(self, name, **kwargs):
        super(MHRRecognizeDatasetsConfig, self).__init__(
        	version=datasets.Version("1.0.0"),
        	name=name,
        	description=_DESCRIPTION,
        	**kwargs,
	    )

class MHRRecognizeDatasets(datasets.GeneratorBasedBuilder):

    BUILDER_CONFIGS = [
        MHRRecognizeDatasetsConfig("whole"),
        MHRRecognizeDatasetsConfig("hole"),
        MHRRecognizeDatasetsConfig("skill"),
    ]

    def _info(self):
        features = datasets.Features({
            "image": datasets.Image(),
            "label": datasets.Value("int32"),
        })
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features
        )

    def _split_generators(self, dl_manager):
        download_files = dl_manager.download_and_extract(_DATA_URL)
        images = dl_manager.iter_files(os.path.join(download_files[self.config.name+"_image"]))
        label = os.path.join(download_files[self.config.name+"_label"])
        return [
            datasets.SplitGenerator(
                name = datasets.Split.TRAIN,
                gen_kwargs = {"images": images, "label": label},
            )
        ]
        
    def _generate_examples(self, images, label):
        label_csv = pd.read_csv(label).fillna(-1)
        for i, path in enumerate(images):
            file_name = os.path.basename(path)
            image_label = label_csv[label_csv["name"] == file_name]['label'].values[0]
            yield i, {"image": path, "label": image_label}