M. I. Mobarakeh

HPMD: A Historical Persian Manuscript Dataset for Word Spotting with Line-Level Annotation

This paper introduces a new dataset of 223 pages, 3,678 lines, 37,631 words, and 130,630 characters, collected from diverse historical Persian books of poetry and prose and annotated at the region, line, and text level and proposes a baseline that is trained only with line-level annotations but returns word-level locat...