A deep learning approach for recognizing ancient Tamil scripts from historical artifacts
A. Umamageswari1, S. Deepa2, L. Sherin Beevi3 and A. Sangari4
Associate Professor, Department of Computer Science and Engineering,SRM Institute of Science and Technology, Ramapuram, Chennai,India1
Assistant Professor, Department of Computer Science and Engineering,SRM Institute of Science and Technology, Ramapuram, Chennai,India2
Assistant Professor, Department of Computer Science and Engineering,R.M.D Engineering College, Kavaraipettai, Tamilnadu,India3
Associate Professor, Department of Electrical and Electronic Engineering,Rajalakshmi Engineering College, Thandalam, Tamilnadu,India4
Assistant Professor, Department of Computer Science and Engineering,SRM Institute of Science and Technology, Ramapuram, Chennai,India2
Assistant Professor, Department of Computer Science and Engineering,R.M.D Engineering College, Kavaraipettai, Tamilnadu,India3
Associate Professor, Department of Electrical and Electronic Engineering,Rajalakshmi Engineering College, Thandalam, Tamilnadu,India4
Corresponding Author : A. Umamageswari
Recieved : 21-February-2024; Revised : 28-April-2025; Accepted : 12-May-2025
Abstract
The preservation and interpretation of ancient scripts are essential for uncovering the rich cultural heritage and historical knowledge of early civilizations. Tamil, one of the world’s oldest languages, contains a vast repository of information preserved in inscriptions, manuscripts, and other historical artifacts. This research proposes a novel methodology for the recognition and deciphering of ancient Tamil words from historical documents and artifacts using image processing techniques and deep learning (DL) algorithms. The proposed framework consists of three main stages: pre-processing, feature extraction, and DL-based recognition. A region-based convolutional neural network (RCNN) architecture is employed to automatically learn and identify the intricate patterns and structural elements of ancient Tamil characters. To enhance feature extraction for irregularly shaped characters, this work introduces Adaptive region of interest (ROI) pooling, which dynamically adjusts to variations in stroke patterns and inscription styles, thereby improving recognition accuracy. The RCNN is trained on a large annotated dataset of ancient Tamil word images, with labels verified by domain experts to ensure data accuracy and reliability. Extensive experiments were conducted on a diverse dataset comprising stone inscriptions, palm leaf manuscripts, and clay tablets. The proposed approach achieved a high recognition accuracy of 98.6%, demonstrating robust performance even under challenging conditions such as stylistic variations, surface degradation, and image noise. This method significantly contributes to the preservation of cultural heritage by enabling the digitization and accessibility of historical texts and inscriptions. By safeguarding valuable linguistic and cultural knowledge, the proposed system ensures its availability for future scholarly research and education.
Keywords
Ancient Tamil, Character recognition, Segmentation, RCNN, Deep learning, Computer vision.
Cite this article
Umamageswari A, Deepa S, Beevi LS, Sangari A. A deep learning approach for recognizing ancient Tamil scripts from historical artifacts. International Journal of Advanced Technology and Engineering Exploration. 2025;12(126):819-832. DOI : 10.19101/IJATEE.2024.111100193
References
[1]
Gautam N, Chai SS. Optical character recognition for Brahmi script using geometric method. Journal of Telecommunication, Electronic and Computer Engineering. 2017; 9(3-11):131-6.
[2]
Bag S, Harit G. A survey on optical character recognition for Bangla and Devanagari scripts. Sadhana. 2013; 38:133-68.
[3]
Kavitha BR, Srimathi CB. Benchmarking on offline handwritten Tamil character recognition using convolutional neural networks. Journal of King Saud University-Computer and Information Sciences. 2022; 34(4):1183-90.
[4]
Ravi J. Handwritten alphabet classification in Tamil language using convolution neural network. International Journal of Cognitive Computing in Engineering. 2024; 5:132-9.
[5]
Sabeenian RS, Paramasivam ME, Anand R, Dinesh PM. Palm-leaf manuscript character recognition and classification using convolutional neural networks. In computing and network sustainability: proceedings of IRSCNS 2018 (pp. 397-404). Springer Singapore.
[6]
Suganya TS, Murugavalli S. Feature selection for an automated ancient Tamil script classification system using machine learning techniques. In international conference on algorithms, methodology, models and applications in emerging technologies 2017 (pp. 1-6). IEEE.
[7]
Dhivya S, Beulah JR. Ancient Tamil character recognition from stone inscriptions–a theoretical analysis. In 2nd Asian conference on innovation in technology 2022 (pp. 1-8). IEEE.
[8]
Subadivya S, Vigneswari J, Yaminie M, Diviya M. Tamilbrahmi script character recognition system using deep learning technique. International Journal of Computer Science and Mobile Computing. 2020; 9(6):114-9.
[9]
Bannigidad P, Sajjan SP. Ancient Kannada handwritten character recognition from palm leaf manuscripts using PyTesseract-OCR technique. In international conference on cognitive computing and information processing 2023 (pp. 154-62). Cham: Springer Nature Switzerland.
[10]
Manigandan TV, Vidhya V, Dhanalakshmi V, Nirmala B. Tamil character recognition from ancient epigraphical inscription using OCR and NLP. In international conference on energy, communication, data analytics and soft computing 2017 (pp. 1008-11). IEEE.
[11]
Rajakumar S, Bharathi VD. 7th century ancient Tamil character recognition from temple wall inscriptions. Indian Journal of Computer Science and Engineering. 2012; 3(5):673-77.
[12]
Vidhyavani A, Manoranjitham T. A survey on recognition of ancient Tamil Brahmi characters from epigraphy. Journal of Pharmaceutical Negative Results. 2022; 13(3):1273-7.
[13]
Vinoth R, Rajesh R, Yoganandhan P. Intelligence system for Tamil vattezhuttuoptical character recognition. International Journal of Computer Science Engineering and Technology. 2017; 8(4):22-6.
[14]
Haritha J, Balamurugan VT, Vairavel KS, Ikram N, Janani M, Indrajith K. CNN based character recognition and classification in Tamil palm leaf manuscripts. In international conference on communication, computing and internet of things 2022 (pp. 1-6). IEEE.
[15]
Subramani K, Subramaniam M. Creation of original Tamil character dataset through segregation of ancient palm leaf manuscripts in medicine. Expert Systems. 2021; 38(1):e12538.
[16]
Islam MA, Iacob IE. Manuscripts character recognition using machine learning and deep learning. Modelling. 2023; 4(2):168-88.
[17]
Sudarsan D, Sankar D. An overview of character recognition from palm leaf manuscripts. In 3rd international conference on smart data intelligence 2023 (pp. 265-72). IEEE.
[18]
Srinivasan R, Babu AK, Balasubramanian PL, Sharma R. Ancient science and Tamil heritage: exploring the interdisciplinary connections for research and revival. In conference proceedings 2023. AIP Publishing LLC.
[19]
Devi PD, Sathiyapriya V. Brahmi script recognition system using deep learning techniques. In third international conference on inventive research in computing applications 2021 (pp. 1346-9). IEEE.
[20]
Durga DK, Uma MP. Recognition and classification of stone inscription character using artificial neural network. In artificial intelligence and technologies: select proceedings of ICRTAC-AIT 2020 (pp. 253-61). Singapore: Springer Singapore.
[21]
Wijerathna KN, Sepalitha R, Indika T, Athauda H, Suranjini PD, Silva JA, et al. Recognition and translation of ancient Brahmi letters using deep learning and NLP. In international conference on advancements in computing 2019 (pp. 226-31). IEEE.
[22]
Karthikeyan S, Jaganathan I, Rameshbabu KS, Abraham A, Gabralla LA, Sivaraj R, et al. Self-adaptive hybridized lion optimization algorithm with transfer learning for ancient Tamil character recognition in stone inscriptions. IEEE Access. 2023; 11:39621-34.
[23]
Magrina MM. Convolution neural network based ancient Tamil character recognition from epigraphical inscriptions. International Research Journal of Engineering and Technology. 2020; 7(04):6130-43.
[24]
Vijayalakshmi R, Gnanasekar JM. A review on character recognition and information retrieval from ancient inscriptions. In 8th international conference on smart structures and systems 2022 (pp.1-7). IEEE.
[25]
Mahajan H, Narayan P, Reddi SI, Sharmila BK. A systematic approach to handle ancient Tamil script using deep learning techniques. In soft computing and signal processing: proceedings of 3rd ICSCSP 2020 (pp. 57-68). Springer Singapore.
[26]
Babu BS, Shetty S, Agarwal A, Sreerama SL, Bhustali VK, Sanjana S. Temple inscriptions recognition and transliteration in Devanagari script. In 2nd international conference on vision towards emerging trends in communication and networking technologies 2023 (pp. 1-6). IEEE.
[27]
Rajakumar S, Bharathi VS. Eighth century Tamil consonants recognition from stone inscriptions. In international conference on recent trends in information technology 2012 (pp. 40-3). IEEE.
[28]
Meng L, Aravinda CV, Uday KR, Izumi T, Yamazaki K. Ancient Asian character recognition for literature preservation and understanding. In Euro-mediterranean conference 2018 (pp. 741-51). Cham: Springer International Publishing.
[29]
Selvapriya B, Rajini V, Sowmya B. Preserving the legacy of ancient Tamil script with deep learning and fuzzy C means algorithm: intelligent approach to digiitization. Journal of Electrical Engineering & Technology. 2025; 20(1):757-71.
[30]
Umamageswari A, Deepa S, Raja K. An enhanced approach for leaf disease identification and classification using deep learning techniques. Measurement: Sensors. 2022; 24:1-6.
[31]
Arasakumaran U, Johnson SD, Sara D, Kothandaraman R. An enhanced identification and classification algorithm for plant leaf diseases based on deep learning. Traitement Du Signal. 2022; 39(3):1013-18.
[32]
Mahajan H, Narayan P, Reddi SI, Sharmila Banu K. A systematic approach to handle ancient Tamil script using deep learning techniques. In soft computing and signal processing: proceedings of 3rd ICSCSP. 2021 (pp. 57-68). Springer Singapore.
[33]
Bhuvaneswari S, Kathiravan K. Enhancing epigraphy: a deep learning approach to recognize and analyze Tamil ancient inscriptions. Neural Computing and Applications. 2024; 36(31):19839-61.
[34]
Hamdany AH, Omar-Nima RR, Albak LH. Translating cuneiform symbols using artificial neural network. TELKOMNIKA (Telecommunication Computing Electronics and Control). 2021; 19(2):438-43.
[35]
Dutta S, Manideep BC, Rai S, Vijayarajan V. A comparative study of deep learning models for medical image classification. In conference series: materials science and engineering 2017 (pp. 1-9). IOP Publishing.
[36]
Brown A. Practical digital preservation: a how-to guide for organizations of any size. Facet Publishing; 2013.
[37]
Richarz J, Vajda S, Grzeszick R, Fink GA. Semi-supervised learning for character recognition in historical archive documents. Pattern Recognition. 2014; 47(3):1011-20.
[38]
Zhu W, Wang X, Li H. Multi-modal deep analysis for multimedia. IEEE Transactions on Circuits and Systems for Video Technology. 2019; 30(10):3740-64.
[39]
Nogales A, Delgado-martos E, Melchor Á, García-tejedor ÁJ. ARQGAN: an evaluation of generative adversarial network approaches for automatic virtual inpainting restoration of Greek temples. Expert Systems with Applications. 2021; 180:115092.
[40]
Yang Z, Yang D, Dyer C, He X, Smola A, Hovy E. Hierarchical attention networks for document classification. In proceedings of the 2016 conference of the North American chapter of the association for computational linguistics: human language technologies 2016 (pp. 1480-9).
[41]
Barucci A, Cucci C, Franci M, Loschiavo M, Argenti F. A deep learning approach to ancient Egyptian hieroglyphs classification. IEEE Access. 2021; 9:123438-47.
[42]
Khayyat MM, Elrefaei LA. Deep reinforcement learning approach for manuscripts image classification and retrieval. Multimedia Tools and Applications. 2022; 81(11):15395-417.
[43]
Gautam N, Dhawas P, Ramteke MA, Madankar M. Application of generative adversarial networks in Brahmi script recognition. In crafting images with generative adversarial networks (GANs) and models 2025 (pp. 221-66). IGI Global Scientific Publishing.
[44]
Bi YQ, Jin JQ, Luo Y, Wang ZY. A robust text recognition system for noisy document images. In proceedings of the international conference on frontiers of artificial intelligence and machine learning 2023 (pp. 19-24).
[45]
Xint Y, Xu L, Guo J, Li J, Sheng X, Zhou Y. Label incorporated graph neural networks for text classification. In 25th international conference on pattern recognition 2021 (pp. 8892-8). IEEE.
[46]
Harish R, Raghavendra Rao GN. Transcription of ancient indian manuscripts through artificial intelligence—current status of technology and the way forward. In international conference on artificial intelligence on textile and apparel 2023 (pp. 339-52). Singapore: Springer Nature Singapore.