Devanagari OCR

The project has source code and data related to the following tools:

1. Optical Character Recognition.
Recognize machine printed Devanagari with or without a dictionary.

2. Document Image Analysis.
Automatic page segmentation of document images in multiple Indian languages. Identifies pictures, lines, and words in a document scanned at 300 dpi.

3. Multi-lingual annotation.
An interface that has transilteration and a soft-keyboard using which multiple languages can be input. The UI also enables users to view the word and character level ground truth of images.


To cite this work, please use:
"Devanagari OCR using a recognition driven segmentation framework and stochastic language models", Suryaprakash Kompalli, Srirangaraj Setlur, Venu Govindaraju, IJDAR, 2009, Volume: 12, Pg.: 123–138