(50-4) 17 * << * >> * Russian * English * Content * All Issues

Classification of Church Slavonic Letter Images from the 10th-14th Centuries Using Neural Networks
A.M. Egorova1, D.V. Demidov1, A.D. Egorov1

1 National Research Nuclear University MEPhI, Kashirskoye Shosse 31, Moscow, 115409, Russia

  Full text (PDF)

DOI: 10.18287/COJ1751

Article ID: 1751

Language: Russian

Abstract:
This paper addresses a problem of recognizing Church Slavonic letters in texts from the 11th-15th centuries using machine learning methods. As part of the study, a database was created containing 20,180 images of 48 different letters, obtained from 647 spreads of liturgical books. To address class imbalance, data augmentation methods were applied.

Various neural network models were tested for letter classification, with EfficientNet-B4 achieving the best results, with F1-weighted scores of 0.976911. However, considering computational efficiency and processing time, an integral quality criterion was calculated, identifying EfficientNet-B1 as the optimal model for practical use. This model processes 43 characters per second, corresponding to 20 seconds per text spread.

The study results demonstrate the feasibility of efficient automated recognition of Church Slavonic texts, significantly accelerating their analysis and processing. Future work will focus on expanding the database and improving image preprocessing methods to enhance recognition accuracy.

Keywords:
Artificial Intelligence, Image Classification, CNN.

Acknowledgements:
This work was financially supported by the Ministry of Science and Higher Education Russian Federation within the "Priority-2030" NRNU MEPhI program.

Citation:
Egorova AM, Demidov DV, Egorov AD. Classification of Church Slavonic Letter Images from the 11th-17th Centuries Using Neural Networks. Computer Optics 2026; 50(4): 1751. doi:10.18287/COJ1751.

References:

  1. Chen X, Jin L, Zhu Y, Luo C, Wang T. Text recognition in the wild: A survey. ACM Computing Surveys 2021; 54(2): 1-35. DOI: 10.1145/3440756.
  2. Singh J, Lehal GS. Comparative performance analysis of feature(s)-classifier combination for Devanagari optical character recognition system. International Journal of Advanced Computer Science and Applications 2014; 5(6): 37-42. DOI: 10.14569/IJACSA.2014.050608.
  3. Smirnova AA, Grafeeva NG, Tokman MA. Recognition of Church Slavonic texts using machine learning methods. Pattern Recognition and Image Analysis 2024; 34(1): 212-218. DOI: 10.1134/S1054661824010209.
  4. Vijayalakshmi R, Gnanasekar JM. A review on character recognition and information retrieval from ancient inscriptions. In: 2022 8th International Conference on Smart Structures and Systems (ICSSS). IEEE; 2022: 1-7. DOI: 10.1109/ICSSS54384.2022.9788041.
  5. Krithiga R, Roja M, Deepa V, Sridevi T, Gowri P. Ancient character recognition: A comprehensive review. IEEE Access 2023. DOI: 10.1109/ACCESS.2023.3264102.
  6. Narang SR, Jindal MK, Kumar M. Ancient text recognition: A review. Artificial Intelligence Review 2020; 53(8): 5517-5558. DOI: 10.1007/s10462-020-09824-5.
  7. Martinovska C, Ivanovski Z, Veleski S, Cepreganova K. Recognition of Old Cyrillic Slavic letters: Decision tree versus fuzzy classifier experiments. In: 2012 6th IEEE International Conference on Intelligent Systems (IS). IEEE; 2012: 48-53. DOI: 10.1109/IS.2012.6335112.
  8. Martinovska Bande C, Veleski S, Cepreganova K, Ivanovski Z. Recognition features for Old Slavic letters: Macedonian versus Bosnian alphabet. International Journal of Scientific and Research Publications 2015; 5(12): 145-153. ISSN 2250-3153.
  9. Rabus A. Recognizing handwritten text in Slavic manuscripts: A neural-network approach using Transkribus. Scripta & e-Scripta 2019; 19: 9-32.
  10. Demidov DV, Lyapin IE, Boriskina EI, et al. Creating a corpus of handwritten heritage of Ancient Rus. In: Proceedings of the International Conference "Corpus Linguistics - 2023". Saint Petersburg: Saint Petersburg State University Press; 2024: 72-81.
  11. Digitized collection of manuscripts and old printed books, GORAZD. Available at: http://gorazd.org/?q=ru/node/106.
  12. Collections, National Library St. Kiril and Metodii. Available at: https://www.nationallibrary.bg/www/our-collections/.
  13. Reul C, Springmann U, Puppe F. LAREX: A semi-automatic open-source tool for layout analysis and region extraction on early printed books. In: Proceedings of the 2nd International Conference on Digital Access to Textual Cultural Heritage. 2017: 137-142. DOI: 10.1145/3078081.3078094.
  14. Sommerschield T, Bodard G, Youn CM, Terras M, Mambrini F, McGillivray B. Machine learning for ancient languages: A survey. Computational Linguistics 2023; 49(3): 703-747. DOI: 10.1162/coli_a_00473.
  15. Chauhan VK, Singh S, Sharma A. HCR-Net: A deep learning based script independent handwritten character recognition network. Multimedia Tools and Applications 2024: 1-35. DOI: 10.1007/s11042-024-19249-x.
  16. Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556; 2014.
  17. Moudgil A, Sharma S, Arora S, Dhir R. Handwritten Devanagari manuscript characters recognition using CapsNet. International Journal of Cognitive Computing in Engineering 2023; 4: 47-54. DOI: 10.1016/j.ijcce.2023.02.002.
  18. Klekovska M, Dimeski K, Najdenov B, Gievska S. Comparison of models for recognition of Old Slavic letters. In: ICT Innovations 2012: Secure and Intelligent Systems. Berlin, Heidelberg: Springer; 2013: 129-139. DOI: 10.1007/978-3-642-37169-1_12.
  19. Kuchuganov A, Lodygyn V, Malevich T, Zolotarev S. Automation of recognizing Old Slavonic manuscripts with help of description logics. Slavistica Vilnensis 2018; 63: 339-352.
  20. Corpus of Handwritten Heritage of Ancient Rus. Available at: https://www.slavcorpora.ru/.
  21. Sokolova M, Japkowicz N, Szpakowicz S. Beyond accuracy, F-score and ROC: A family of discriminant measures for performance evaluation. In: Australasian Joint Conference on Artificial Intelligence. Berlin, Heidelberg: Springer; 2006: 1015-1021. DOI: 10.1007/11941439_114.
  22. Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, Kaiser Ł, Polosukhin I. Attention is all you need. In: Advances in Neural Information Processing Systems 30 (NeurIPS 2017); 2017. arXiv:1706.03762.
  23. LeCun Y, Bottou L, Bengio Y, Haffner P. Gradient-based learning applied to document recognition. Proceedings of the IEEE 1998; 86(11): 2278-2324. DOI: 10.1109/5.726791.
  24. Jamil S, Jalil Piran M, Kwon OJ. A comprehensive survey of transformers for computer vision. Drones 2023; 7(5): 287. DOI: 10.3390/drones7050287.
  25. Keles FD, Wijewardena PM, Hegde C. On the computational complexity of self-attention. In: International Conference on Algorithmic Learning Theory (ALT 2023). Proceedings of Machine Learning Research; 2023: 597-619.
  26. Pan X, Zhan X, Dai B, Lin D, Loy CC. On the integration of self-attention and convolution. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR); 2022: 815-825. DOI: 10.1109/CVPR52688.2022.00086.
  27. He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2016: 770-778. DOI: 10.1109/CVPR.2016.90.
  28. Tan M, Le Q. EfficientNet: Rethinking model scaling for convolutional neural networks. In: Proceedings of the 36th International Conference on Machine Learning (ICML). Proceedings of Machine Learning Research; 2019: 6105-6114.
  29. Xie S, Girshick R, Dollár P, Tu Z, He K. Aggregated residual transformations for deep neural networks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2017: 1492-1500. DOI: 10.1109/CVPR.2017.634.
  30. Loshchilov I, Hutter F. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101; 2017. DOI: 10.48550/arXiv.1711.05101.

151, Molodogvardeiskaya str., Samara, 443001, Russia; E-mail: journal@computeroptics.ru; Tel: +7 (846) 242-41-24 (Executive secretary), +7 (846) 332-56-22 (Issuing editor), Fax: +7 (846) 332-56-20