About QA dataset (csv) format
Opened this issue · 1 comments
jinulee-v commented
Hello,
Thanks for such an interesting dataset.
I do not need the whole set of scanned images, but only the textual information.
Is there a way to get the OCR-ed results and key-value pairs in CSV? I assume the output of dataset_scripts/get_dataset_qa.sh is what I am looking for.
The aws s3 sync
download is much more massive than I thought, and I hope there is a workaround for it.
Best regards,
Jinu
benleetownsend commented
Apologies for the delayed response. You should just be able to download using the following link patterns
https://s3.us-east-2.wasabisys.com/project-fruitfly/{dataset}/{split}.csv
For splits: train, test, val
and datasets: fcc_invoices, s1_pages, nda, charities, resource_contracts