Bengali Handwritten Alphabet Recognition using Deep Convolutional Neural Network
Sifat Ahmed, Fatima Tabsun, Abdus Sayef Reyadh, Asif Imtiaz Shaafi, Faisal Muhammad Shah · 2019 International Conference on Computer, Communication, Chemical, Materials and Electronic Engineering (IC4ME2) · 2019
The technology of object recognition has advanced in recent times. There are so many models that performed brilliantly when used to detect English handwritten alphanumeric characters. But when it comes to Bengali handwritten character recognition, these predefined model slightly underperformed. The complexity of Bengali characters and unavailability of a good dataset are the main reasons for the underperformance of these models. So the problem still quite unsolved. But in Bengali digit recognition some of the models performed very well. We propose a Deep Convolutional Neural Network model to recognize Bengali handwritten alphabets. The dataset used in this experiment is pretty new and have not tested with DCNN models. Till now, our model achieves 95% accuracy in recognizing the alphabets.