OCR0065: Create repo on Hugging face for all the datasets we have of OCR

[ta4tsering]

Description:
So currently we have a lot of OCR data that we have annotated and all of those images are on s3 with each image as a single object and also the cvs files are all in s3. So to make the datasets easily usable and easily accessible I will creating the zip files with all the images and upload to the hugging face repo with the transcriptions and all with the data split or data distribution that eric used.

Completion Criteria:

All the Tibetan OCR data uploaded to Openpecha hugging face.

Subtasks:

• Lhasa Kanjur
• Lithang Kanjur
• Derge Tenjur
• Norbuketaka
• Google Books
• Betsug data
• Durtsa data
• update the script for the special case of google books and norbuketaka data

note:
for the Norbuketaka and Google books, we already have a hugging face repo but without the data distributions so I am using that hugging face repo to create the new hugging face repo on Openpecha hugging face with the data distribution but without the zipped image file

Card Reviewer:

• @10kalden

[ta4tsering]

https://huggingface.co/datasets/openpecha/OCR-Lhasakanjur

[kaldan007]

@ta4tsering kindly reach out to @gangagyatso4364 regarding how to combine the uchen dataset in one hugging face dataset.

[kaldan007]

Kindly add url to the dataset

[ta4tsering]

https://huggingface.co/datasets/openpecha/OCR-Durtsa
https://huggingface.co/datasets/openpecha/OCR-Betsug
https://huggingface.co/datasets/openpecha/OCR-Lithangkanjur
https://huggingface.co/datasets/openpecha/OCR-Google_Books

[ta4tsering]

https://huggingface.co/datasets/openpecha/OCR-Norbuketaka
above is the norbuketaka hugging face repo and below is the missing images from the data.
norbuketaka_missing.txt
and below is the data distribution used by eric.
norbuketaka data distribution

[ta4tsering]

I wasnt able to do for the Derge Tenjur, it is taking way too long to fix the issue which is that the images arent present in the zip as it is compromised when downloaded from the hugging face and since it is not on the s3 I cant use the url as well.