Downloading the dataset
Hi!
I would like to download the LC25000 dataset from here using git clone git@hf.co:datasets/1aurent/LC25000 as suggested in https://huggingface.co/docs/hub/en/datasets-downloading.
But I am not quite sure how to deal with the four parquet files LC25000/data/train-*.parquet since I do not see the images.
Can you please provide the instructions for downloading this data?
Kind regards,
George
Hi @GeorgeBatch !
The dataset was uploaded to the huggingface hub using the datasets library, because of this all the images and labels were encoded in the parquet/arrow format.
You can't easily access the images from this format, the simplest way use the data in this format is to use the datasets library, as instructed slightly above in the docs you linked:
from datasets import load_dataset
ds = load_dataset("1aurent/LC25000")
You could always extract the images from the dataset using the ds object this way, though it would probably be easier to download the original dataset from the academictorrent (see https://github.com/tampapath/lung_colon_image_set).
Hi @1aurent , thank you for the explanation!