Funded through grants from The Andrew W. Mellon Foundation, the Open Islamicate Texts Initiative Arabic-script OCR Catalyst Project (OpenITI AOCP) is the first undertaking of its kind to tackle the technical and organizational barriers that historically have stymied the development of Arabic-script OCR and digital text production for Islamicate Studies.
Two phases of the project have been funded until now:
- OpenITI AOCP Phase 1 (2019-2021): The primary technical goal of the first phase of OpenITI AOCP was to achieve ≥97% character accuracy rates (CARs) for OCR on the most used Persian and Arabic print typefaces. Read more
- OpenITI AOCP Phase 2 (2022-2025): The aims of the second phase of OpenITI AOCP include expanding the size of OpenITI’s Persian and Arabic corpus through large-scale OCR work in Persian and Arabic; extending the linguistic capabilities of OpenITI’s OCR tools into Ottoman Turkish and Urdu; and transforming its open-source optical character recognition (OCR)/handwritten text recognition (HTR) pipeline by incorporating newly developed unsupervised machine learning tools into its workflow. Read more
OpenITI AOCP invested in expanding Scripta’s open OCR platform eScriptorium; the OpenITI instance of eScriptorium can be found here.
Segmentation and transcription models trained for Arabic script in the context of this project can be found here. These are Kraken models, which can be used in eScriptorium or via the command line.