FamilySearch Line-Level Dataset of Transcribed Handwritten Ottoman Nüfūs Records

We present a line-level annotated dataset of historical Ottoman Nüfūs records.The transcriptions for these images can be used by the public for creating handwritten text recognition (HTR) systems.The lines are transcribed in their original script, and accompanying images have been preprocessed for use in HTR.Each line has been either transcribed or reviewed by an expert human linguist to ensure the accuracy and quality of the data.Furthermore, we present a smaller set of labels to seed automated transliteration efforts.