Handwritten Amharic Character Recognition Through Transfer Learning: Integrating CNN Models and Machine Learning Classifiers

Handwritten Amharic character recognition presents significant challenges due to the script’s syllabic nature and variations in handwriting styles. This study investigates a hybrid approach that integrates convolutional neural networks (CNNs) with machine learning classifiers to enhance r...

Full description

Saved in:
Bibliographic Details
Main Authors: Natenaile Asmamaw Shiferaw, Zefree Lazarus Mayaluri, Prabodh Kumar Sahoo, Ganapati Panda, Prince Jain, Adyasha Rath, Md. Shabiul Islam, Mohammad Tariqul Islam
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10935359/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850046872881725440
author Natenaile Asmamaw Shiferaw
Zefree Lazarus Mayaluri
Prabodh Kumar Sahoo
Ganapati Panda
Prince Jain
Adyasha Rath
Md. Shabiul Islam
Mohammad Tariqul Islam
author_facet Natenaile Asmamaw Shiferaw
Zefree Lazarus Mayaluri
Prabodh Kumar Sahoo
Ganapati Panda
Prince Jain
Adyasha Rath
Md. Shabiul Islam
Mohammad Tariqul Islam
author_sort Natenaile Asmamaw Shiferaw
collection DOAJ
description Handwritten Amharic character recognition presents significant challenges due to the script’s syllabic nature and variations in handwriting styles. This study investigates a hybrid approach that integrates convolutional neural networks (CNNs) with machine learning classifiers to enhance recognition accuracy. Transfer learning is applied using four CNN architectures: AlexNet, VGG16, VGG19, and ResNet50 as feature extractors. Initially, their performance is evaluated with softmax classifiers. Subsequently, the softmax layer is replaced with machine learning classifiers, including Random Forest, XGBoost, and Support Vector Machine (SVM), while freezing the pretrained feature extractors. The Hybrid ResNet50 + SVM model achieves the highest accuracy of 91.89%, with a precision of 92.46%, recall of 91.15%, and an F1-score of 91.80%. These results indicate that SVM serves as a potential alternative to softmax, offering robust classification performance for complex handwritten scripts. This research contributes to advancements in handwritten character recognition systems for underrepresented languages.
format Article
id doaj-art-64a9dfb7212f4e8f8c3022aceefd3e09
institution DOAJ
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-64a9dfb7212f4e8f8c3022aceefd3e092025-08-20T02:54:22ZengIEEEIEEE Access2169-35362025-01-0113521345214810.1109/ACCESS.2025.355319910935359Handwritten Amharic Character Recognition Through Transfer Learning: Integrating CNN Models and Machine Learning ClassifiersNatenaile Asmamaw Shiferaw0https://orcid.org/0009-0002-3243-3401Zefree Lazarus Mayaluri1https://orcid.org/0000-0002-6910-126XPrabodh Kumar Sahoo2https://orcid.org/0000-0003-0112-7129Ganapati Panda3https://orcid.org/0000-0002-3555-5685Prince Jain4https://orcid.org/0000-0002-7950-7263Adyasha Rath5Md. Shabiul Islam6https://orcid.org/0000-0002-4630-7249Mohammad Tariqul Islam7https://orcid.org/0000-0002-4929-3209Department of Computer Science and Engineering, C. V. Raman Global University, Bhubaneswar, Odisha, IndiaDepartment of Electrical Engineering, C. V. Raman Global University, Bhubaneswar, Odisha, IndiaDepartment of Mechatronics Engineering, Parul Institute of Technology, Parul University, Vadodara, Gujarat, IndiaDepartment of Electronics and Communication Engineering, C. V. Raman Global University, Bhubaneswar, Odisha, IndiaDepartment of Mechatronics Engineering, Parul Institute of Technology, Parul University, Vadodara, Gujarat, IndiaDepartment of Computer Science and Engineering, C. V. Raman Global University, Bhubaneswar, Odisha, IndiaFaculty of Engineering, Multimedia University, Cyberjaya, Selangor, MalaysiaDepartment of Electrical, Electronic and Systems Engineering, Faculty of Engineering and Built Environment, UKM Bangi, Selangor, MalaysiaHandwritten Amharic character recognition presents significant challenges due to the script’s syllabic nature and variations in handwriting styles. This study investigates a hybrid approach that integrates convolutional neural networks (CNNs) with machine learning classifiers to enhance recognition accuracy. Transfer learning is applied using four CNN architectures: AlexNet, VGG16, VGG19, and ResNet50 as feature extractors. Initially, their performance is evaluated with softmax classifiers. Subsequently, the softmax layer is replaced with machine learning classifiers, including Random Forest, XGBoost, and Support Vector Machine (SVM), while freezing the pretrained feature extractors. The Hybrid ResNet50 + SVM model achieves the highest accuracy of 91.89%, with a precision of 92.46%, recall of 91.15%, and an F1-score of 91.80%. These results indicate that SVM serves as a potential alternative to softmax, offering robust classification performance for complex handwritten scripts. This research contributes to advancements in handwritten character recognition systems for underrepresented languages.https://ieeexplore.ieee.org/document/10935359/Amharic scripthandwritten character recognitiondeep learningconvolutional neural networks (CNN)transfer learning
spellingShingle Natenaile Asmamaw Shiferaw
Zefree Lazarus Mayaluri
Prabodh Kumar Sahoo
Ganapati Panda
Prince Jain
Adyasha Rath
Md. Shabiul Islam
Mohammad Tariqul Islam
Handwritten Amharic Character Recognition Through Transfer Learning: Integrating CNN Models and Machine Learning Classifiers
IEEE Access
Amharic script
handwritten character recognition
deep learning
convolutional neural networks (CNN)
transfer learning
title Handwritten Amharic Character Recognition Through Transfer Learning: Integrating CNN Models and Machine Learning Classifiers
title_full Handwritten Amharic Character Recognition Through Transfer Learning: Integrating CNN Models and Machine Learning Classifiers
title_fullStr Handwritten Amharic Character Recognition Through Transfer Learning: Integrating CNN Models and Machine Learning Classifiers
title_full_unstemmed Handwritten Amharic Character Recognition Through Transfer Learning: Integrating CNN Models and Machine Learning Classifiers
title_short Handwritten Amharic Character Recognition Through Transfer Learning: Integrating CNN Models and Machine Learning Classifiers
title_sort handwritten amharic character recognition through transfer learning integrating cnn models and machine learning classifiers
topic Amharic script
handwritten character recognition
deep learning
convolutional neural networks (CNN)
transfer learning
url https://ieeexplore.ieee.org/document/10935359/
work_keys_str_mv AT natenaileasmamawshiferaw handwrittenamhariccharacterrecognitionthroughtransferlearningintegratingcnnmodelsandmachinelearningclassifiers
AT zefreelazarusmayaluri handwrittenamhariccharacterrecognitionthroughtransferlearningintegratingcnnmodelsandmachinelearningclassifiers
AT prabodhkumarsahoo handwrittenamhariccharacterrecognitionthroughtransferlearningintegratingcnnmodelsandmachinelearningclassifiers
AT ganapatipanda handwrittenamhariccharacterrecognitionthroughtransferlearningintegratingcnnmodelsandmachinelearningclassifiers
AT princejain handwrittenamhariccharacterrecognitionthroughtransferlearningintegratingcnnmodelsandmachinelearningclassifiers
AT adyasharath handwrittenamhariccharacterrecognitionthroughtransferlearningintegratingcnnmodelsandmachinelearningclassifiers
AT mdshabiulislam handwrittenamhariccharacterrecognitionthroughtransferlearningintegratingcnnmodelsandmachinelearningclassifiers
AT mohammadtariqulislam handwrittenamhariccharacterrecognitionthroughtransferlearningintegratingcnnmodelsandmachinelearningclassifiers