High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

Keyword Recognition by Improving Recurrent Neural Network using Character Model

Author(s):

Geetanjali Bhagwani , L.J. INSTITUTE OF ENGINEERING AND TECHNOLOGY; Ompriya Kale, L.J. INSTITUTE OF ENGINEERING AND TECHNOLOGY

Keywords:

Handwritten recognition, keyword spotting, bidirectional long short-term memory, recurrent neural network, CTC token passing algorithm, Character model

Abstract

Keyword spotting refers to the process of retrieving all instances of a given keyword from these documents. In the present paper, a keyword spotting method for handwritten documents is obtained using different various systems for offline handwriting recognition. A new technique is used for robust keyword spotting that uses bidirectional Long Short-Term Memory (BLSTM) recurrent neural nets and CTC Token Passing Algorithm to incorporate contextual information in documents. The document analysis is done by demonstrating tri-gram character model to significantly improve the spotting performance on IAM offline database.

Other Details

Paper ID: IJSRDV3I40681
Published in: Volume : 3, Issue : 4
Publication Date: 01/07/2015
Page(s): 1306-1309

Article Preview

Download Article