eScriptorium: Digital Text Production for Urdu, Hindi, and Bengali Print, part 3
OCR of historical printing in Bengali using segmentation and recognition models trained in Kraken from an annotated dataset of Bengali texts published between 1860 and 1940.