Work

Improving the Eastern Arabic Hand Written Digits Recognition Using Deep Learning

Year: 2025

Type: article

Abstract: Convolutional neural networks CNNs or simply deep learning was designed for the challenging of object recognition such as digits, characters and even objects that can be classified. The deep convolutional networks are designed to be trained through using a large dataset of the classes. The degree of complexity of recognizing handwritten digits are different from person to person and even from language to language. The Eastern Arabic handwritten digits recognition system (EAH), which has ten classes, is one of the most challenging object recognitions due to its complexity and similarities of classes. There are many techniques worked on proposed systems for recognizing EAH but with limited accuracy and output cost function. In this paper, a proposed new technique with multisided layers is utilized for Eastern Arabic handwritten recognition. This technique proved to have higher output accuracy and performance than other existing systems.

Source: Journal of Computer Science

DOI: 10.3844/jcssp.2025.1293.1298

Authors: Hamdy Morsy, Hamdy Morsy, Hashim M. Hasanien, Mostafa M. Abutaleb, Mostafa M. Abutaleb

Institutions: Qassim University, Helwan University

Language: English

Topic: Handwritten Text Recognition Techniques

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funder: Qassim University

Award: G92618698