Arabic Handwriting Classification using Deep Transfer Learning Techniques

Almisreb, Ali Abd; Tahir, Nooritawati Md; Turaev, Sherzod; Junid, Syed Abdul Mutalib Al

doi:10.47836/pjst.30.1.35

Cited by 11 publications

(6 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This network has 19 layers and was trained on one million photos in 1000 categories from the ImageNet database. These 19 layers consist of 16 convolutional, three fully connected CNN with stride and padding of 1, and 2*2 max pooling layers (Almisreb, Turaev, Saleh, & Al Junid, 2022). This network only contains 3x3 convolutional layers stacked on top of one another to increase depth.…”

Section: Vgg19mentioning

confidence: 99%

Handwritten Character Recognition in Assyrian Language Using Convolutional Neural Network

Armya,

Abdulrazzaq

2024

SJUOZ

View full text Add to dashboard Cite

Academics and researchers worldwide have paid close attention to biometric handwriting recognition using deep learning as much research has been proposed to enhance biometric recognition in the past and in recent years. Several solutions for character recognition systems in various languages, including Chinese, English, Japanese, Arabic, and Kurdish have been developed. Unfortunately, there has been minimal growth in the Assyrian language. There is still little research on Assyrian handwriting. In this paper, a new Assyrian language dataset was created as part of the procedure by distributing 500 forms consisting of 36 Assyrian characters to people between the ages of 13 and 60 of both genders. The preprocessing operation includes cleaning the noisy data and segmenting each image to 224x224 pixels. This effort resulted in the collection of 18,000 images of these characters to be trained 70% and tested 30% in four CNN models, VGG16, VGG19, MobileNet-V2, and ResNet-50, over 30 epochs to give an accuracy rate of 90.97%, 92.06%, 95.70%, and 94.97%., respectively.

show abstract

Section: Vgg19mentioning

confidence: 99%

Handwritten Character Recognition in Assyrian Language Using Convolutional Neural Network

Armya,

Abdulrazzaq

2024

SJUOZ

View full text Add to dashboard Cite

show abstract

“…Almisreb, et al [17] assess and validate the usage of deep transfer learning models. Hench, the best model for categorizing handwritten photographs written by native or non-native, author is determined using seven different deep learning transfer models, including AlexNet, GoogleNet, ResNet18, ResNet50, ResNet101, VGG16, and VGG19.…”

Section: Related Workmentioning

confidence: 99%

Comparison of CNN’s Architecture GoogleNet, AlexNet, VGG-16, Lenet -5, Resnet-50 in Arabic Handwriting Pattern Recognition

Nugraha

Darmawan

Dwiyansaputra

2023

KINETIK

View full text Add to dashboard Cite

The Arabic script is written from right to left and consists of 28 characters, with no capital or lowercase letters. The Arabic script has several orthographic and morphological properties that make handwriting recognition of the Arabic script challenging. In addition, one of the biggest challenges in recognizing Arabic script patterns is the different handwriting styles and characters of each person's writing. The authors propose a study to compare the accuracy of handwriting pattern recognition in Arabic script which has been done previously by comparing five CNN architectures, namely GoogleNet, AlexNet, VGG-16, LeNet-5, and ResNet-50. Considering that previous research has not obtained excellent accuracy. The number of datasets used is 8400 image data and the most optimal comparison of testing and training data is 80:20. Based on the research that has been done, there are several things that the author can conclude. The model is made using 64 filters for each convolution layer because the optimal size is used for 5 architectures, kernel size is 3x3, neurons is 128, dropout weight is 50% to reduce overfitting, learning rate is 0.001, image size is 64x64, the normalization method with the ReLU activation function, and 1-dimensional input image (grayscale), and with a comparison of testing and training data of 80:20. The VGG-16 architectural model is the architecture that gets the highest score, namely 83.99%. This can have good potential to be developed as a medium for learning Arabic script.

show abstract

“…Comparison of CNN architectures such as VGG-16, VGG-19, ResNet-18, ResNet-50, ResNet 101, AlexNet, and GoogleNet in the Arabic handwritten image classification shows that GoogleNet has a higher accuracy value than other CNN architectures. The classification system generated by the GoogleNet architecture achieves the highest accuracy of 95.5% [24] . Based on previous research, the GoogleNet architecture has a good ability in image classification.…”

Section: Previous Researchesmentioning

confidence: 99%

Ultrasound Image Synthetic Generating Using Deep Convolution Generative Adversarial Network For Breast Cancer Identification

Haq

Fatichah

2023

JTS

View full text Add to dashboard Cite

Breast cancer is the leading cause of death in women worldwide; prevention of possible death from breast cancer can be decreased by early identification ultrasound image analysis by classifying ultrasound images into three classes (Normal, Benign, and Malignant), where the dataset used has imbalanced data. Imbalanced data cause the classification system only to recognize the majority class, so it is necessary to handle imbalanced data. In this study, imbalanced data can be handled by implementing the Deep Convolution Generative Adversarial Network (DCGAN) method as the addition of synthetic images to the training data. The DCGAN method generates synthetic images with feature learning on a Convolutional Neural Network (CNN), making DCGAN more stable than the basic generative adversarial network method.Synthetic and original images were further classified using the CNN GoogleNet method, which performs well in image classification and with reasonable computation cost. Synthetic ultrasound images were generated using a tuning hyperparameter in the DCGAN method to adjust the input size on GoogleNet for imbalanced data handling. From the experiment result, the implementation of DCGAN-GoogleNet has a higher accuracy in handling imbalanced data than conventional augmentation and other previous research, with an accuracy value reaching 91.61%, which is 1% to 4% higher than the accuracy value in the previous method.

show abstract

Arabic Handwriting Classification using Deep Transfer Learning Techniques

Cited by 11 publications

References 21 publications

Handwritten Character Recognition in Assyrian Language Using Convolutional Neural Network

Handwritten Character Recognition in Assyrian Language Using Convolutional Neural Network

Comparison of CNN’s Architecture GoogleNet, AlexNet, VGG-16, Lenet -5, Resnet-50 in Arabic Handwriting Pattern Recognition

Ultrasound Image Synthetic Generating Using Deep Convolution Generative Adversarial Network For Breast Cancer Identification

Contact Info

Product

Resources

About