Title: Offline Arabic handwritten character recognition: from conventional machine learning system to deep learning approaches

Authors: Soumia Faouci; Djamel Gaceb; Mohammed Haddad

Addresses: Laboratoire d'Informatique, de Modélisation, d'Optimisation et des Systèmes Électroniques (LIMOSE), Faculté des Sciences, Université M'Hamed Bougara, Boumerdès, Algeria ' Laboratoire d'Informatique, de Modélisation, d'Optimisation et des Systèmes Électroniques (LIMOSE), Faculté des Sciences, Université M'Hamed Bougara, Boumerdès, Algeria ' Lab LIRIS, UMR CNRS 5205, University of Claude Bernard Lyon 1, F69622, Villeurbanne, France

Abstract: Researchers have made great strides in the area of Arabic handwritten character recognition in the last decades especially with the fast development of deep learning algorithms. The characteristics of Arabic manuscript text pose several problems for a recognition system. This paper presents a conventional machine learning system based on the extraction of a set of preselected features and an SVM classifier. In the second part, a simplified convolutional neural network (CNN) model is proposed, which is compared to six other CNN models based on the pre-trained architectures. The suggested methods were tested using three databases: two versions of the OIHACDB dataset and the AIA9K dataset. The experimental results show that the proposed CNN model obtained promising results, as it is able to recognise 94.7%, 98.3%, and 95.6% of the test set of the three databases OIHACDB-28, OIHACDB-40, and AIA9K, respectively.

Keywords: deep learning; DL; convolutional neural network; CNN; Arabic handwritten character recognition; machine learning; support vector machines; SVM; transfer learning; features extractor; FE; fine-tuning; FT; pre-trained model; conventional machine learning system.

DOI: 10.1504/IJCSE.2022.124562

International Journal of Computational Science and Engineering, 2022 Vol.25 No.4, pp.385 - 398

Received: 01 Apr 2021
Accepted: 16 Sep 2021
Published online: 28 Jul 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article