Khmer handwritten digits recognition using convolution neural networks| International Journal of Innovative Science and Research Technology

Khmer Handwritten Digits Recognition using Convolution Neural Networks

Authors : Ly Sreypov

Volume/Issue : Volume 10 - 2025, Issue 7 - July

Google Scholar : https://tinyurl.com/ju9zs8vk

DOI : https://doi.org/10.38124/ijisrt/25jul531

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Note : Google Scholar may take 30 to 40 days to display the article.

Abstract : This study focuses on developing a Convolutional Neural Network (CNN) model to recognize and classify handwritten Khmer numbers from a dataset of 19,530 images. The research addresses the challenge of duplicated number recognition by leveraging CNNs, which are highly effective for image recognition tasks in computer vision. The dataset is preprocessed, cleaned, scaled, and split into training, validation, and testing sets. Using libraries such as NumPy, Pandas, TensorFlow, Keras, and Scikit-learn, a CNN model is constructed, trained, and evaluated, achieving a 95% accuracy in predicting handwritten Khmer numbers from 0 to 9. The work highlights the efficiency and robustness of CNNs compared to other networks for this task, contributing to improved handwritten number recognition.

Keywords : Machine Learning, Handwritten Digits Recognition, Convolution Neural Network (CNN).

References :

Sarayut Gonwirat and Olarik Surinta ,Improving Recognition of Thai Handwritten Characters with deepconvolutional neural network, India, 20Aprill 2020.
Bayram Annanurov1,2 and Norliza Mohd Noor1, "KHMER HANDWRITTEN TEXT RECOGNITION WITH CONVOLUTION NEURAL NETWORKS," ARPN Journal of Engineering and Applied Sciences , pp. 8828-8833, 2018 .
Xuchen Song, Xue Goa, Yanfanf Ding and zhixin Wang, A handwritten Chinese characters recognition method based on sample set expansion and CNN, Shanghai, China: IEEE, 2016 3^rd International Conference on Systems and Informatics (ICSAI).

This study focuses on developing a Convolutional Neural Network (CNN) model to recognize and classify handwritten Khmer numbers from a dataset of 19,530 images. The research addresses the challenge of duplicated number recognition by leveraging CNNs, which are highly effective for image recognition tasks in computer vision. The dataset is preprocessed, cleaned, scaled, and split into training, validation, and testing sets. Using libraries such as NumPy, Pandas, TensorFlow, Keras, and Scikit-learn, a CNN model is constructed, trained, and evaluated, achieving a 95% accuracy in predicting handwritten Khmer numbers from 0 to 9. The work highlights the efficiency and robustness of CNNs compared to other networks for this task, contributing to improved handwritten number recognition.

Keywords : Machine Learning, Handwritten Digits Recognition, Convolution Neural Network (CNN).