Recognition of printed Devanagari text using BLSTM Neural Network

Author

Sankaran, Naveen ; Jawahar, C.V.

Author_Institution

Int. Inst. of Inf. Technol., Hyderabad, India

fYear

2012

fDate

11-15 Nov. 2012

Firstpage

322

Lastpage

325

Abstract

In this paper, we propose a recognition scheme for the Indian script of Devanagari. Recognition accuracy of Devanagari script is not yet comparable to its Roman counterparts. This is mainly due to the complexity of the script, writing style etc. Our solution uses a Recurrent Neural Network known as Bidirectional LongShort Term Memory (BLSTM). Our approach does not require word to character segmentation, which is one of the most common reason for high word error rate. We report a reduction of more than 20% in word error rate and over 9% reduction in character error rate while comparing with the best available OCR system.

Keywords

natural language processing; optical character recognition; recurrent neural nets; text analysis; BLSTM neural network; Devanagari script recognition accuracy; Indian Devanagari script; OCR system; Roman counterparts; bidirectional long-short term memory; character segmentation; printed Devanagari text recognition; recurrent neural network; script complexity; word error rate; writing style; Accuracy; Character recognition; Degradation; Feature extraction; Logic gates; Neural networks; Optical character recognition software; BLSTM; Devanagari; OCR; Word recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Pattern Recognition (ICPR), 2012 21st International Conference on

Conference_Location

Tsukuba

ISSN

1051-4651

Print_ISBN

978-1-4673-2216-4

Type

conf

Filename

6460137