This competition to transcribe digitised historic Arabic manuscripts might be of interest.
The British Library has a growing collection of digitised Arabic manuscripts but traditional OCR techniques can struggle to work with these materials making it difficult for the library and researchers to work with these collections in a machine-readable format. Researchers have tried a number of different machine learning approaches to tackle this challenge and have launched competitions to help encourage new approaches.
You can find a description of the results of a previous competition in this paper
More details on the competition and entry information can be found on the ICDAR2019 Competition on Recognition of Historical Arabic Scientific Manuscripts page.