Akshara: Contextual Transliteration of Brahmi Script Using Deep Learning

View at publisher Summarise with VIRAT
Authors: Utsav Khairnar, Vivek Suvarna, Niranjan Pillai, Harshit Bhat, Sangeetha Selvan
Cited by: 0
References: 16

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Recognizing Brahmi script is a big challenge in Optical Character Recognition (OCR) and a key part of understanding historical documents. Working with old scripts unfolds all kinds of problems: faded manuscripts, messy handwriting, and characters that don’t look anything like each other. The usual OCR tricks, like template matching or rule-based systems, just don’t cut it for Brahmi. But deep learning has changed the game. Now we have tools like Transformers, CNNs, and RNNs that push recognition much further than before. Still, there’s a catch: most modern models need a ton of labeled data, and for something as old as Brahmi, that’s hard to get. So, in this study, we built a hybrid deep learning model that combines CNNs and RNNs, and we leaned on data augmentation and transfer learning to boost accuracy without needing huge datasets. Along with explaining how we put together our dataset, trained the model, and measured its performance, this paper compares the already existing methods to solve the problem and our approach. Our goal is pretty straightforward: make Brahmi recognition easier and more accurate, and in the process, help preserve historical texts. That way, researchers in linguistics and archaeology can get to these materials without jumping through so many hoops

Fields of Study & Subjects

Handwritten Text Recognition Techniques Vehicle License Plate Recognition Advanced Neural Network Applications

Publication details

Journal / source
Biennial International Conference on Nascent Technologies in Engineering (ICNTE)
Pages
1–6
Publication date
16 January 2026
Language
EN