Recognition and Classification of Handwritten Urdu Numerals Using Deep Learning Techniques
Urdu is a complex language as it is an amalgam of many South Asian and East Asian languages; hence, its character recognition is a huge and difficult task. It is a bidirectional language with its numerals written from left to right while script is written in opposite direction which induces complexi...
Main Authors: | , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2023-01-01
|
Series: | Applied Sciences |
Subjects: | |
Online Access: | https://www.mdpi.com/2076-3417/13/3/1624 |
_version_ | 1797625077118795776 |
---|---|
author | Aamna Bhatti Ameera Arif Waqar Khalid Baber Khan Ahmad Ali Shehzad Khalid Atiq ur Rehman |
author_facet | Aamna Bhatti Ameera Arif Waqar Khalid Baber Khan Ahmad Ali Shehzad Khalid Atiq ur Rehman |
author_sort | Aamna Bhatti |
collection | DOAJ |
description | Urdu is a complex language as it is an amalgam of many South Asian and East Asian languages; hence, its character recognition is a huge and difficult task. It is a bidirectional language with its numerals written from left to right while script is written in opposite direction which induces complexities in the recognition process. This paper presents the recognition and classification of a novel Urdu numeral dataset using convolutional neural network (CNN) and its variants. We propose custom CNN model to extract features which are used by Softmax activation function and support vector machine (SVM) classifier. We compare it with GoogLeNet and the residual network (ResNet) in terms of performance. Our proposed CNN gives an accuracy of 98.41% with the Softmax classifier and 99.0% with the SVM classifier. For GoogLeNet, we achieve an accuracy of 95.61% and 96.4% on ResNet. Moreover, we develop datasets for handwritten Urdu numbers and numbers of Pakistani currency to incorporate real-life problems. Our models achieve best accuracies as compared to previous models in the literature for optical character recognition (OCR). |
first_indexed | 2024-03-11T09:51:43Z |
format | Article |
id | doaj.art-13ee1bea14054342905626092592a02d |
institution | Directory Open Access Journal |
issn | 2076-3417 |
language | English |
last_indexed | 2024-03-11T09:51:43Z |
publishDate | 2023-01-01 |
publisher | MDPI AG |
record_format | Article |
series | Applied Sciences |
spelling | doaj.art-13ee1bea14054342905626092592a02d2023-11-16T16:08:10ZengMDPI AGApplied Sciences2076-34172023-01-01133162410.3390/app13031624Recognition and Classification of Handwritten Urdu Numerals Using Deep Learning TechniquesAamna Bhatti0Ameera Arif1Waqar Khalid2Baber Khan3Ahmad Ali4Shehzad Khalid5Atiq ur Rehman6School of Electrical Engineering and Computer Science, National University of Sciences and Technology, Islamabad 24090, PakistanSchool of Electrical Engineering and Computer Science, National University of Sciences and Technology, Islamabad 24090, PakistanComputer Engineering Department, Bahria University, Islamabad 44000, PakistanDepartment of Electrical and Computer Engineering, International Islamic University, Islamabad 04436, PakistanDepartment of Software Engineering, Bahria University, Islamabad 44000, PakistanComputer Engineering Department, Bahria University, Islamabad 44000, PakistanArtificial Intelligence and Intelligent Systems Research Group, School of Innovation, Design and Engineering, Mälardalen University, Högskoleplan 1, 722 20 Västerås, SwedenUrdu is a complex language as it is an amalgam of many South Asian and East Asian languages; hence, its character recognition is a huge and difficult task. It is a bidirectional language with its numerals written from left to right while script is written in opposite direction which induces complexities in the recognition process. This paper presents the recognition and classification of a novel Urdu numeral dataset using convolutional neural network (CNN) and its variants. We propose custom CNN model to extract features which are used by Softmax activation function and support vector machine (SVM) classifier. We compare it with GoogLeNet and the residual network (ResNet) in terms of performance. Our proposed CNN gives an accuracy of 98.41% with the Softmax classifier and 99.0% with the SVM classifier. For GoogLeNet, we achieve an accuracy of 95.61% and 96.4% on ResNet. Moreover, we develop datasets for handwritten Urdu numbers and numbers of Pakistani currency to incorporate real-life problems. Our models achieve best accuracies as compared to previous models in the literature for optical character recognition (OCR).https://www.mdpi.com/2076-3417/13/3/1624urdu numeral recognitionconvolutional neural networkSVMGoogLeNetResNet |
spellingShingle | Aamna Bhatti Ameera Arif Waqar Khalid Baber Khan Ahmad Ali Shehzad Khalid Atiq ur Rehman Recognition and Classification of Handwritten Urdu Numerals Using Deep Learning Techniques Applied Sciences urdu numeral recognition convolutional neural network SVM GoogLeNet ResNet |
title | Recognition and Classification of Handwritten Urdu Numerals Using Deep Learning Techniques |
title_full | Recognition and Classification of Handwritten Urdu Numerals Using Deep Learning Techniques |
title_fullStr | Recognition and Classification of Handwritten Urdu Numerals Using Deep Learning Techniques |
title_full_unstemmed | Recognition and Classification of Handwritten Urdu Numerals Using Deep Learning Techniques |
title_short | Recognition and Classification of Handwritten Urdu Numerals Using Deep Learning Techniques |
title_sort | recognition and classification of handwritten urdu numerals using deep learning techniques |
topic | urdu numeral recognition convolutional neural network SVM GoogLeNet ResNet |
url | https://www.mdpi.com/2076-3417/13/3/1624 |
work_keys_str_mv | AT aamnabhatti recognitionandclassificationofhandwrittenurdunumeralsusingdeeplearningtechniques AT ameeraarif recognitionandclassificationofhandwrittenurdunumeralsusingdeeplearningtechniques AT waqarkhalid recognitionandclassificationofhandwrittenurdunumeralsusingdeeplearningtechniques AT baberkhan recognitionandclassificationofhandwrittenurdunumeralsusingdeeplearningtechniques AT ahmadali recognitionandclassificationofhandwrittenurdunumeralsusingdeeplearningtechniques AT shehzadkhalid recognitionandclassificationofhandwrittenurdunumeralsusingdeeplearningtechniques AT atiqurrehman recognitionandclassificationofhandwrittenurdunumeralsusingdeeplearningtechniques |