import datasets import json _CITATION = "" _DESCRIPTION = "Dataset for training agents with Maze-v0 environment." _HOMEPAGE = "https://huggingface.co/datasets/NathanGavenski/imagetest" _LICENSE = "" _REPO = "https://huggingface.co/datasets/NathanGavenski/imagetest" class ImageSet(datasets.GeneratorBasedBuilder): def _info(self): return datasets.DatasetInfo( description=_DESCRIPTION, features=datasets.Features({ "obs": datasets.Image(), "actions": datasets.Value("int32"), "rewards": datasets.Value("float32"), "episode_starts": datasets.Value("bool"), "maze": datasets.Value("string"), "image": datasets.Image() }), supervised_keys=None, homepage=_HOMEPAGE, citation=_CITATION, ) def _split_generators(self, dl_manager): image_path = dl_manager.download_and_extract(f"{_REPO}/resolve/main/images.tar.gz") info_path = dl_manager.download_and_extract(f"{_REPO}/resolve/main/dataset.tar.gz") return [ datasets.SplitGenerator( name=datasets.Split.TRAIN, gen_kwargs={ "images": f"{image_path}/images", "infos": info_path } ), ] def _generate_examples(self, images, infos): print(images) with open(f"{infos}/dataset.jsonl", encoding="utf-8") as data: for idx, line in enumerate(data): record = json.loads(line) image = record["obs"].split(".")[0] yield idx, { "obs": } # for idx, ((filepath, image), info) in enumerate(zip(images, infos)): # for idx, info in enumerate(infos): # print(idx, info) # yield idx, { # # "image": {"path": filepath, "bytes": image.read()}, # }