MatriVasha

Abstract

Bengali or Bangla language is native to Bengal. Bengali speaks their mother tongue, which they feel comfortable. In 1952, UNESCO announced that 21st February is the international mother language day. People remember this day with great respect. We achieved our mother languages for our martyrs. And, there are belongs to a different type of characters. Bangla’s handwritten compound character is one of them. For helping Bengali people in different purposes, Bangla’s handwritten compound character recognition performs a vital role. Such as Bangla character recognition, they are extracting data from handwritten paper and digitalization of any kind of documents, etc. In Bangla language, there are have ten numerical digits, 50 basic characters, greater than 200 compound characters, and ten modifiers. But there is no strong dataset for Bangla character recognition. Most of the researchers are working for only digit or basic characters or a few numbers of compound characters. But works with compound characters are complicated because of their different shapes and styles. For this kind of reason, MatriVasha has proposed a dataset designed to recognize 120 compound characters in Bangla that consist of 306,240‬ images written by 2552 distinct writers from Bangladesh. Our proposed dataset is the most comprehensive dataset for handwritten compound characters in Bangla; so far, MatriVasha datasets can be found on this link https://shahariarrabby.github.io/matrivasha.