License and commercial-use clarification for Arabic OCR training

#2
by ahmed-sayed-Ocr - opened

Hello,

I am working on an Arabic OCR research and development project and would like to use the "mssqpi/Arabic-OCR-Dataset", including its approximately 2.16 million labeled image-text samples.

The dataset page currently does not appear to specify a license. Could you please confirm whether the complete dataset may be used for the following purposes?

  1. Training and fine-tuning Arabic OCR models.
  2. Research and evaluation.
  3. Commercial use of the resulting models or services.
  4. Modifying, augmenting, rendering, or applying synthetic degradation to the samples.
  5. Publishing trained model weights, checkpoints, adapters, or LoRA weights.
  6. Creating and distributing derived datasets without redistributing the original samples unchanged.

Could you also clarify:

  • The license that applies to the dataset.
  • The original sources of the images and text.
  • Whether all samples are created or owned by you, or whether any samples originate from third-party datasets.
  • The required attribution or citation.

A clear written confirmation here would allow us to record the permission with the dataset documentation.

Thank you.

Sign up or log in to comment