Proceedings of the 1st International Conference on Artificial Intelligence, Communication, IoT, Data Engineering and Security, IACIDS 2023, 23-25 November 2023, Lavasa, Pune, India

Research Article

Deciphering Ancient Inscriptions with Optical Character Recognition

Download56 downloads
  • @INPROCEEDINGS{10.4108/eai.23-11-2023.2343235,
        author={Anitha  Julian and Devipriya  R},
        title={ Deciphering Ancient Inscriptions with Optical Character Recognition},
        proceedings={Proceedings of the 1st International Conference on Artificial Intelligence, Communication, IoT, Data Engineering and Security, IACIDS 2023, 23-25 November 2023, Lavasa, Pune, India},
        publisher={EAI},
        proceedings_a={IACIDS},
        year={2024},
        month={3},
        keywords={deciphering inscriptions cnn ocr character recognition},
        doi={10.4108/eai.23-11-2023.2343235}
    }
    
  • Anitha Julian
    Devipriya R
    Year: 2024
    Deciphering Ancient Inscriptions with Optical Character Recognition
    IACIDS
    EAI
    DOI: 10.4108/eai.23-11-2023.2343235
Anitha Julian1,*, Devipriya R1
  • 1: Saveetha Engineering College, Chennai, India
*Contact email: cse.anithajulian@gmail.com

Abstract

Archaeologists strive to gain a deeper understanding of historical contexts across various regions by deciphering ancient Tamil inscriptions. However, this manual decoding process demands a considerable amount of labor and time. The inefficiency of this traditional approach has the potential to hinder future archaeological research. To address this issue, a proposed effort is underway to create an Optical Character Recognition (OCR) system specifically designed for the interpretation of medieval Tamil inscriptions. This study primarily focuses on the OCR module, which combines fully convolutional technologies with adaptive neuro-fuzzy inference (ANN). By conducting tests using actual images, the recognition rates of these two micro technologies were compared for the initial training set, preprocessed test data, and test data. Ultimately, the CNN-based OCR module emerged as the superior solution for this purpose.