Arabic OCR
Arabic is hard to read by machine: connected letterforms, four shapes per letter, optional diacritics, right-to-left flow, and documents that mix Arabic with English, numbers, and tables. Qomrah's OCR is built for exactly this — not adapted from a Latin engine as an afterthought.
You get clean, searchable text with preserved reading order, layout coordinates, and per-region confidence — ready to index, store, or feed into the next stage.
- Printed and handwritten Arabic
- Diacritics, ligatures, and stamps
- Multi-column and tabular layouts
- Confidence scores per region