Arabic Handwritten Text Recognition Dataset from the Muharaf Corpus

A re-uploaded version of the Muharaf dataset created by Mehreen Saeed et al. for research purposes. The dataset is hosted on Hugging Face for easier access and experimentation, with a last recorded update on 2025-12-31.

Use Cases

• Train Arabic handwritten text recognition models based on the image data.
• Benchmark OCR algorithms for Arabic script using the provided handwritten samples.
• Study the variation in Arabic handwriting styles and character shapes.

Strengths

• Dataset is derived from a known research corpus (Muharaf) created by Mehreen Saeed et al.
• Hosted on Hugging Face for streamlined access and integration into machine learning workflows.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset scale are unknown, which may limit suitability assessment.

Provenance

Source

Original Muharaf dataset by Mehreen Saeed et al.

Collection Method

Re-uploaded to Hugging Face for accessibility.

Time Range

null

Freshness

Last updated 2025-12-31 19:19:26; freshness should be verified.

Geography

null