Bangla Handwritten Compound Character Recognition

Abstract

Developing new algorithms or findings related to any language requires a rich dataset. In Bangla language, most of the datasets are consist of 50 basic characters and a few compound characters. Though many researchers have worked on Bangla printed and handwritten characters but still, the dataset of most compound characters (handwritten) is not done yet. In this paper, a dataset of 76 handwritten compound characters and 3 modifiers are preprocessed. All those characters are unique and absent in other datasets of Bangla characters. The dataset is available publicly at [https://bit.ly/2PBEIrj].