randall-lab/arabic-characters

Dataset Card for Arabic Characters

Dataset Details

Dataset Description

This dataset contains 16,800 Arabic handwritten characters, written by 60 participants. It is intended for Arabic character recognition tasks using machine learning. The dataset is split into a training set of 13,440 images and a test set of 3,360 images, with 28 Arabic characters (labeled 0–27). Each image is 32×32 pixels in grayscale, scanned at 300 dpi and preprocessed. The original source is the Arabic Handwritten Characters Dataset.

• License: Open Database License (ODbL)

Dataset Sources

• Homepage: https://github.com/mloey/Arabic-Handwritten-Characters-Dataset
• Paper: El-Sawy, A., Loey, M., & El-Bakry, H. (2017). Arabic handwritten characters recognition using convolutional neural network. WSEAS Transactions on Computer Research, 5(1), 11-19.

Dataset Structure

Total images: 16,800

Splits:

• Train: 13,440 images (80%)

• Test: 3,360 images (20%)

Classes (labels): 28 (Arabic letters), labeled 0–27

Image specs: PNG format, 32×32 pixels, grayscale

Example Usage

Below is a quick example of how to load this dataset via the Hugging Face Datasets library.

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("randall-lab/arabic-characters", split="train", trust_remote_code=True)
# dataset = load_dataset("randall-lab/arabic-characters", split="test", trust_remote_code=True)

# Access a sample from the training set
example = dataset["train"][0]
image = example["image"]
label = example["label"]

image.show()  # Display the image
print(f"Label: {label}")

Citation

BibTeX:

@article{el2017arabic, title={Arabic handwritten characters recognition using convolutional neural network}, author={El-Sawy, Ahmed and Loey, Mohamed and El-Bakry, Hazem}, journal={WSEAS Transactions on Computer Research}, volume={5}, pages={11--19}, year={2017} }