Recognition of Arabic Air-Written Letters - Machine Learning, Convolutional Neural Networks, and Optical Character Recognition (OCR) Techniques - Article

Aug 5, 2026

Upload by Aria

Air writing has become a key interaction paradigm that can support communication between humans and intelligent systems, including metaverse-oriented scenarios. Prior research mainly covers English and Chinese, leaving limited work for Arabic. This study proposes a hybrid approach combining feature extraction and deep learning with machine learning and OCR, using grid and random search to optimize model parameters. Deep CNN features (VGG16, VGG19, SqueezeNet) train models on the AHAWP dataset, with preprocessing to reduce bias and OCR integration to segment letters; the best reported accuracy reaches 88.8%.

LanguageEnglish

CategoryTechnology

FormatPDF

Pages27

Reading Time1 h 8 min

AccessFree to download

Introduction

Proposed hybrid approach

Feature extraction with deep CNNs

Machine learning classifiers

OCR-based letter isolation

Dataset, preprocessing, and optimization

Results and accuracy

Texas A&M University-San Antonio Digital Commons @ Texas A&M University-San Antonio All Faculty Scholarship 11-28-2023 Recognition of Arabic Air-Written Letters: Machine Learning, Convolutional Neural Networks, and Optical Character Recognition (OCR) Techniques Khalid Nahar Izzat Alsmadi Rabia Emhamed Al Mamlook Ahmad Nasayreh Hasan Gharaibeh See next page for additional authors Follow this and additional works at: [https://digitalcommons.tamusa.edu/pubs_faculty](https://digitalcommons.tamusa.edu/pubs_faculty) Part of the Computer Sciences Commons Authors Khalid Nahar, Izzat Alsmadi, Rabia Emhamed Al Mamlook, Ahmad Nasayreh, Hasan Gharaibeh, Ali Saeed Almuflih, and Fahad Alasim Article Recognition of Arabic Air-Written Letters: Machine Learning, Convolutional Neural Networks, and Optical Character Recognition (OCR) Techniques Khalid M. O. Nahar, Izzat Alsmadi, Rabia Emhamed Al Mamlook , Ahmad Nasayreh, Hasan Gharaibeh, Ali Saeed Almuflih and Fahad Alasim [https://doi.org/10.3390/s23239475](https://doi.org/10.3390/s23239475) sensors Article Recognition of Arabic Air-Written Letters: Machine Learning, Convolutional Neural Networks, and Optical Character Recognition (OCR) Techniques Khalid M. O. Nahar 1, *, Izzat Alsmadi 2, Rabia Emhamed Al Mamlook 3,4,*, Ahmad Nasayreh 1, Hasan Gharaibeh 1, Ali Saeed Almu􀀃ih 5 and Fahad Alasim 6 Citation: Nahar, K.M.O.; Alsmadi, I.; Al Mamlook, R.E.; Nasayreh, A.; Gharaibeh, H.; Almu􀀃ih, A.S.; Alasim, F. Recognition of Arabic Air-Written Letters: Machine Learning, Convolutional Neural Networks, and Optical Character Recognition (OCR) Techniques. Sensors 2023, 23, 9475. [https://](https://)[ ](https://)[doi.org/10.3390/s23239475](doi.org/10.3390/s23239475) [Received: 18 September 2023](Received: 18 September 2023) Revised: 15 November 2023 Accepted: 18 November 2023 Published: 28 November 2023 Copyright: © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://](https://)[ ](https://)[creativecommons.org/licenses/by/](creativecommons.org/licenses/by/)[ ](creativecommons.org/licenses/by/)[4.0/](4.0/)) . 1 Computer Science Department, Faculty of Information Technology and Computer Sciences, Yarmouk University, Irbid 21163, Jordan; [nasayrahahmad@gmail.com](nasayrahahmad@gmail.com) (A.N.); [hasangharaibeh87@gmail.com](hasangharaibeh87@gmail.com) (H.G.) 2 Department of Computing and Cyber Security, Texas A&M University-San Antonio, San Antonio, TX 78224, USA; [izzat.alsmadi@tamusa.edu](izzat.alsmadi@tamusa.edu) 3 Department of Business Administration, Trine University, Angola, IN 49008, USA 4 Department of Mechanical and Industrial Engineering, University of Zawia, Tripoli 16418, Libya 5 Department of Industrial Engineering, College of Engineering, King Khalid University, Abha 62529, Saudi Arabia; asalmu􀀃ih@kku.edu.sa 6 Department of Industrial Engineering, College of Engineering, King Saud University, Riyadh 11495, Saudi Arabia; [falasim@ksu.edu.sa](falasim@ksu.edu.sa) * Correspondence: [khalids@yu.edu.jo](khalids@yu.edu.jo) (K.M.O.N.); [almamlookr@trine.edu](almamlookr@trine.edu) (R.E.A.M.) Abstract: Air writing is one of the essential 􀀂elds that the world is turning to, which can bene􀀂t from the world of the metaverse, as well as the ease of communication between humans and machines. The research literature on air writing and its applications shows signi􀀂cant work in English and Chinese, while little research is conducted in other languages, such as Arabic. To 􀀂ll this gap, we propose a hybrid model that combines feature extraction with deep learning models and then uses machine learning (ML) and optical character recognition (OCR) methods and applies grid and random search optimization algorithms to obtain the best model parameters and outcomes. Several machine learning methods (e.g

FAQs

What problem does the paper address?

It addresses recognition of Arabic letters written in the air, where gesture variability and lack of dedicated Arabic research make recognition challenging.

How does the proposed method work?

It extracts deep features from CNNs, trains machine learning classifiers, applies preprocessing to improve data quality, and integrates OCR to isolate individual letters from continuous gestures.

Which dataset is used for training and evaluation?

The study uses the AHAWP dataset, which includes diverse writing styles and hand sign variations.

Aug 5, 2026

Upload by Aria