Hugging Face's logo

Datasets:
hy270
/
ICT_Books

Tasks:
Image-to-Text
Document Question Answering
Visual Question Answering
Languages:
English
Urdu
Arabic
Size:
10K<n<100K
Tags:
education
textbooks
ocr
page-truth
pakistan
nbf
License:
Dataset card Files
xet
 Community
1
ICT_Books / USAGE.md
hy270's picture
hy270
Add files using upload-large-folder tool
8b890f6 verified
preview code
|
Raw History Blame Contribute Delete
1.28 kB

Usage & IP note

This dataset is PRIVATE and RESTRICTED to internal Rumi/Orenda use.

Source & copyright

The page images and their transcriptions are derived from National Book Foundation (NBF) and FBISE textbooks (grades 6–12), which are copyrighted by their respective publishers. A subset of pages (English textbook literary anthology) contains third-party copyrighted literary works (novel and play excerpts).

Permitted

  • Internal use by the Rumi/Orenda team to build and improve the lesson-plan pipeline: retrieval / embeddings, seed-image reuse, layout/OCR research, evaluation.

Not permitted

  • Public release, redistribution, or resale of this dataset or any substantial extract.
  • Making the Hugging Face repo public, or removing its access controls, while it contains verbatim copyrighted textbook/literary content.
  • Using the copyrighted literary-excerpt pages (source=ocr-fallback) in any teacher-facing output without a separate rights review.

Before teacher-facing use

Filter or human-verify needs_human_review pages (religious/Arabic content and all OCR-fallback pages) — these are lower-fidelity by design and flagged accordingly.

If in doubt about a specific use, ask the FEAT-080 owner before proceeding.