Advancements in Offline Handwriting-Based Language Recognition: A Comprehensive Review for Indic Scripts

Text recognition is essential to Natural Language Processing (NLP) and the digitization process. These systems play a crucial role in addressing the limitations of non-editable documents in the process of digitalization. Additionally, they have significant applications in various fields, including the financial sector, the medical field, the field of machine translation, digital libraries, and many more. Furthermore, due to the pandemic, there has been a surge in the volume of digital data within the education industry, hence requiring the implementation of text recognition systems to manage it effectively. Text recognition systems were designed to process three distinct forms of text: (a) machine-printed text, (b) offline handwritten text, and (c) online handwritten text. This review focuses primarily on exploring the functionality of Offline Handwritten Text (OHT) Recognition systems. The analysis revealed 23 articles that focused only on the recognition of Indian language text within the timeframe spanning from 2018 to 2023. Convergence has a significant obstacle in the form of limited access to historical documents and other conventional resources across various text formats. Although this research investigates all Indian languages, the Hindi language/Devanagari script is the primary emphasis. This study comprehensively examines all existing text recognition methods for Indian scripts. Furthermore, the assessment of textual recognition work across multiple Indian languages is conducted using various metrics, i.e., accuracy, precision, Fmeasure, etc.