Deep Learning Based Ahom Alphabet Recognition System Using Convolutional Neural Network

DOI

No DOI

Published In

Volume 5, Issue 2 (2026)

Published Online

May-August 2026

Pages

721-730

Article Type

Original Article

Check for updates

No DOI available for cross-reference

The preservation of ancient and low-resource languages has become an important area of research in Artificial Intelligence and Natural Language Processing. Ahom, an ancient Tai language historically used in Assam, India, possesses significant cultural and historical importance. However, due to limited digital resources and the absence of automated recognition systems, the preservation and computational processing of Ahom script remain challenging. This paper presents a Deep Learning based Ahom Alphabet Recognition System capable of recognizing handwritten Ahom characters using Convolutional Neural Networks (CNN). The proposed system accepts both uploaded images and real-time drawing input through an interactive web interface. The input image undergoes preprocessing techniques including grayscale conversion, resizing, and normalization before being passed into the CNN model for classification. The system is implemented using Python, TensorFlow, Flask, HTML, CSS, and JavaScript. Hugging face is utilized for public deployment and remote accessibility. Experimental results demonstrate that the CNN-based model effectively recognizes Ahom alphabets with promising accuracy. The system contributes toward digital preservation of the Ahom script and provides a foundation for future research in low-resource script recognition, Optical Character Recognition (OCR), and AI-driven heritage preservation.