A CNN-Based Intelligent Framework for Modi Script Character Recognition

In this paper, a Modi Script Character Recognition and Translation System has been presented to address the challenge of preserving and accessing historical Modi manuscripts. The system utilizes Optical Character Recognition (OCR) along with deep learning techniques, particularly Convolutional Neural Networks (CNN), to accurately recognize characters from scanned images.   The recognized text is further transliterated into Devanagari and translated into modern languages such as Marathi, Hindi, and English, making it accessible to a wider audience. Additionally, the integration of Text-to-Speech (TTS) enhances usability by providing audio output.   The proposed system demonstrates effective performance in recognizing and translating Modi script, thereby contributing to cultural preservation and digital accessibility. Future work can focus on improving accuracy, expanding the dataset, and developing mobile-based applications for wider usability.

1. U. Pal, B.B. Chaudhuri, Indian script character recognition: a survey, Pattern Recognit. 37 (2004) 1887–1899. https://doi.org/10.1016/j.patcog.2004.02. 003.   2. S. Joseph, J. George, Feature extraction and classification techniques of MODI script character recognition, Pertanika J. Sci. Technol. 27 (2019) 1649–1669.   3. R. Khillari, History of MODI Script, Shri Modi Vibhav. (2008). http://modiscript.blogspot.com/2008/05/history-of-modi-script.html. 4. R. Thakre, Reviving MODI Script, Typogr. 2014. 1 (2014). http://shivajiscript .blogspot.com (accessed December 5, 2018). 5. R.A. Ramraje, History of MODI Script in Maharashtra, Glob. Online Electron. Int. Interdiscip. Res. J. ( GOEIIRJ). 2 (2013). 6. S. Kulkarni, P. Borde, R. Manza, P. Yannawar, Text Line Segmentation of Handwritten Historical MODI Documents, in: Int. Conf. Information, Commun. Comput. Technol., 2017: pp. 1–12.