A Simplistic & Fast approach to Bangla Handwritten Digit Recognition using Convolutional Neural Network

Abstract

Handwritten digit recognition has consistently a major test because of its variety of shape, size, and composing style.Accurate Handwritten Digit Recognition is becoming challenging and thoughtful to researchers due to its educational and economic values.Most of the To benchmark Bengali digit acknowledgment calculations, a huge openly accessible dataset is required which is liberated from inclinations starting from topographical area, sexual orientation, and age.In light of this point, NumtaDB, a dataset comprising of something else than 85,000 pictures of transcribed Bengali digits, has been amassed.The challenges of NumtaDB data set is that it contains unbiased, unprocessed and augmented images.So for this reason different kinds preprocessing steps were followed to process the available data and A simplistic fast approach to Bangla Handwritten digit recognition using Convolution Neural Network is proposed.