Persian Handwritten Digits and Symbols (PHDS) Image Processing Dataset

Published: 15 September 2025| Version 1 | DOI: 10.17632/2kh93j3vgn.1
Contributors:
, Pouya Rostami, Mahdi Zare,

Description

The dataset structure is designed to include two main directories, "train" and "test." Each of these directories is divided into subfolders, where data is organized based on the type of character (handwritten digits from 0 to 9 and mathematical symbols such as parentheses, brackets, braces, and arithmetic signs). The images, after digitization, have been resized to a standard dimension of 28x28 pixels, and various filters (including Gaussian Blur, Median Blur, and Canny) were applied for noise reduction and enhancement of clarity. This preprocessing ensures data uniformity, enabling direct use in deep learning and computer vision algorithms.

Files

Institutions

  • University of Zanjan Faculty of Sciences
    Zanjan, Zanjan

Categories

Artificial Intelligence, Computer Vision, Artificial Neural Network, Image Processing, Applied Computing, Handwriting Recognition, Deep Learning

Licence