A two phase trained Convolutional Neural Network for Handwritten Bangla Compound Character Recognition

Recognizing Bangla compound characters is a challenging problem due to its high curly nature. In this paper, we propose a convolutional neural network (CNN) architecture to recognize handwritten Bangla compound characters. The learning of proposed architecture is done in two phase. In the first phase, a CNN is trained in an unsupervised way to minimize the reconstruction loss. Afterward, these weights are used to initialize the starting layers of second CNN to reduce the recognition loss through supervised learning. The effectiveness of the proposed model is validated on compound character dataset CMATERdb 3.1.3.3, which consists of 171 different character classes. It achieves recognition results of 93.90% and 97.37 % in top 1 and top 2 choices. The recognition performance outperforms state-of-the-art method for handwritten Bangla compound characters by a margin of 3.57%.

I. Introduction

BangIa is an Indo-Aryan language and one of the highly spoken languages of India and an official language of Bangladesh. 250 million peoples of the world speak Bangla language and make it as the seventh largest spoken language in the world. Thus, developing an OCR for handwritten Bangla language is an important problem. However, a good recognition accuracy is being achieved for handwritten English characters [1]. But, for the Indic language such as Bangla, character recognition is still not fully solved and ample of scope remains to improve the existing state-of-the-art methods.