Datasets:

Modalities:
Text
Formats:
csv
ArXiv:
Libraries:
Datasets
pandas
License:

How to decompress deid_png.part* files for the images?

#2
by LostGeorge - opened

Hi, not sure if I am missing something obvious, but I'm seeing that the dataset partitions deid_png.part00, deid_png.part01, etc. are z-standard compressed files. But when I try to decompress with zstd, I get Read error(39): premature end.

I also tried just loading in with datasets.load_dataset, and if I just pass the repository, the dataset is only a json dict dataset for the corresponding csv/json metadata files. If I try to pass the deid_png files as the data_files arg, it fails saying there are no files with a supported extension. Is there an example of how to load in the images?

This is with datasets==3.3.1 on ubuntu 20.04 if that matters.

Rajpurkar Lab / Harvard Medical AI org
edited May 5

I split the large archive file into several parts. To access the PNG images from these part files, you can:

  • Concatenate the files into a single archive: cat deid_png.part* > deid_png.tar
  • Extract the archive: tar -xvf deid_png.tar

This should properly reconstruct the archive and extract the images. Please let me know if there are any questions. Thank you!

LostGeorge changed discussion status to closed

Sign up or log in to comment