[Note: To use this dataset, cite the article: GreekNet: Handwritten Greek Alphabet Recognition Using Explainable Parallel CNN with Attention Mechanisms] This dataset is a curated collection of handwritten Greek alphabets, designed to facilitate research and development in handwriting recognition, image classification, and related machine learning tasks. It contains images of all 24 letters of the Greek alphabet, collected from various handwritten forms and processed into structured subsets for easy use in training and testing models. Folder Structure: All Files in Forms:This folder contains the original forms from which the handwritten Greek alphabet samples were collected. Each form contains multiple instances of handwritten letters in various styles. Cropped Original Files:Handwritten alphabet samples cropped directly from the original forms. Each file represents a single character, retaining the unique handwriting style of the contributor. Working Dataset_Split (70-15-15): Contains the dataset split into training (70%), validation (15%), and testing (15%) subsets. The training set has been augmented to balance class distribution and ensure robust model training. The validation and test sets are unaltered and serve for model evaluation and testing. Key Features: Covers all 24 letters of the Greek alphabet in handwritten form. Provides original forms for reference, along with cropped alphabet samples for precise image-based analysis. Balanced and augmented training set for improved model performance. Ideal for applications in explainable AI (XAI), character recognition, and linguistic research. Intended Use:The dataset is suitable for developing machine learning and deep learning models for handwritten character recognition. It can also be used in studies focusing on the interpretability of AI models, human handwriting variability, and cross-linguistic pattern analysis. License: Creative Commons Attribution 4.0 International.This dataset is publicly available and can be used for research and educational purposes. Please cite appropriately if used in publications. How to Access:doi.org/10.5281/zenodo.14554030