Handwritten Bangla Numeral Recognition using Convolutional Neural Networks

Handwritten Bangla numeral recognition has been studied broadly in the last century. This is a very interest subject among the experimenters because of the progression of various pattern recognition algorithms. In this paper, we compare the results of most widely used machine learning method Like Mu...

Full description

Saved in:
Bibliographic Details
Published in2018 2nd International Conference on Electronics, Materials Engineering and Nano Technology (IEMENTech) pp. 1 - 4
Main Authors Paul, Jaya, Sarkar, Anasua
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.05.2018
Subjects
Character recognition
Convolution
Convolutional neural networks
Feature extraction
Handwriting recognition
Training
Online AccessGet full text
DOI10.1109/IEMENTECH.2018.8465296

Cover

More Information
Summary:Handwritten Bangla numeral recognition has been studied broadly in the last century. This is a very interest subject among the experimenters because of the progression of various pattern recognition algorithms. In this paper, we compare the results of most widely used machine learning method Like Multi Layer Percepton (MLP) and deep learning method like multilayer Convolutional Neural Network (CNN) to get the accuracy of 99.2% using CNN as compared to 97.97% using MLP on Bangla numeral image database named Halder et al.
DOI:10.1109/IEMENTECH.2018.8465296