Recognize Arabic handwritten using CNN model

Abstract EN

One of the most challenges that face machine learning is handwritten recognition, especially Arabic scripts, because many styles found for Arabic font.

in this paper, an investigation model is proposed to make recognition for Arabic handwritten scripts utilizing convolutional neural network (CNN), with multi layers of normalization and regularization to reduce training time and increase overall accuracy, with validation accuracy 98% for kaggle dataset for Arabic handwritten characters and digits using python.