Submit Your Manuscript Online

Offline Bangla Handwritten Character Recognition with Convolutional Neural Network (CNN)

	International Journal of Scientific Research and Engineering Development (IJSRED)	Full Text:
	Published Issue : Volume-4 Issue-1
	Year of Publication : 2021
	Unique Identification Number : IJSRED-V4I1P120
	Authors : Md. Zahidul Islam, Md. Abdul Based, Md.Mahbubur Rahman
	: Click Here | DOI : Update Soon

Abstract :

Now-a-days Handwritten Character Recognition (HCR) is a remarkable research topic in Bangla language. It is known to be one of the most first-rate classical problems in the field of machine learning and image processing. HCR has been researched extensively during the last few decades with varying level of success. It is so much challenging motive owing to its high variation in individual writing style and structural reflection between characters. Significantly due to lack of large Bangla handwritten character dataset, Bangla handwritten character recognition could not program far. In this field, several types of character recognition policies are going on, however offline handwritten Bangla documents are rarely found. Several approaches such as neural network, handcraft feature, support vector machine, deep learning have employed for handwritten character recognition. In this paper, the proposed approach has been evaluated by using Convolutional Neural Netwok (CNN) and greedy algorithm to recognize Bangla handwritten character and achieve better recognition accuracy. The used dataset are comparatively large and reliable for character recognition. The proposed method gained more than 90% recognition accuracy which is significantly better approach.