Tamil Digitization Project
Name of the Researcher(s)
- Dr. AR Weerasinghe – UCSC
- Mr. Chamila Liyanage – UCSC
- Dr. Thilini Nadungodage – UCSC
- Mr. Viraj Welgama – UCSC
- Dr. EYA Charles – University of Jaffna
- Dr. S Mahesan – University of Jaffna
- Dr. A Ramanan – University of Jaffna
Description
The main goal of this project was to digitize important Tamil literature and other documents that are out of print. Under this project criteria for selecting books which need to be digitized was defined prioritized. In order to define an OCR alphabet for Tamil, sample data for common type faces were collected. These were then used to train a Tamil optical character recognition (OCR) model in a way that the model gives the optimum performance for a diverse set of font types and sizes. The project was to develop the tools needed to automatically recognize the most common printed Tamil fonts from scanned images of books and documents for digitizing such content. Currently there are no freely available accurate OCR solutions for printed Tamil scripts globally. As such, the baseline is that any scanned or hardcopy printed Tamil artifacts that need to be digitized are manually re?typed on word processing software. This project aimed to provide an OCR solution for Tamil in which the level of accuracy that vastly reduces the time needed to digitize printed Tamil text. The OCR developed is to be used as part of a pipeline for converting historically important Tamil literature to digital form and passed on to proof-reading and collation processes via a crowd sourcing effort through an online interface. For this purpose, we used the Noolaham project for scanning Tamil literature as a partner and handed over the subtask of creating the online interface to the University of Jaffna.
Related Images
Contact details of the researcher(s)
- Dr. A.R.Weerasinghe Email: arw@ucsc.cmb.ac.lk
- Mr. Chamila Liyanage – UCSC Email: cml@ucsc.cmb.ac.lk
- Dr. Thilini Nadungodage – UCSC Email: hnd@ucsc.cmb.ac.lk
- Mr. Viraj Welgama – UCSC Email: wvw@ucsc.cmb.ac.lk
Publication
- Chamila Liyanage, Thilini Nadungodage, Ruvan Weerasinghe (2015). Developing a commercial grade Tamil OCR for recognizing font and size independent text. International Conference on Advances in ICT for Emerging Regions. Colombo, Sri Lanka, 24-25 August 2015.
(ocr.tamil.subasa.lk)