A Benchmark Dataset of Sinhala Handwritten Text for Optical Character Recognition (OCR) System Development

Fuente: Zenodo
Gespeichert in:
Bibliographische Detailangaben
1. Verfasser: Rathnayaka, Maheesha Sewmini
Format: Recurso digital
Sprache:Englisch
Veröffentlicht: Zenodo 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866902214450085888
author Rathnayaka, Maheesha Sewmini
author_facet Rathnayaka, Maheesha Sewmini
contents <p>This dataset contains a collection of images representing Sinhala handwritten characters, words, and paragraphs, covering all characters and connectors. It includes 25,516 scanned images gathered from 381 participants, representing diverse handwriting styles across different demographic groups.</p> <p>This dataset was developed as part of a research project on Sinhala handwritten text recognition.</p> <p>The dataset is annotated with labels at the character, word, and paragraph levels, and includes linked metadata (age, gender, education/occupation category, residence type, and dominant hand) using anonymized participant IDs. Demographic information is recorded separately in an Excel file (<code>MetaData.xlsx</code>) to ensure anonymity.</p> <p>Images were scanned at 300 DPI and split into training, validation, and test sets. Data augmentation was applied only to the training dataset to improve model generalization. The <code>Raw</code> folder contains the original training, validation, and test datasets, while the <code>Augmented</code> folder contains the augmented version of the training dataset.</p> <p>A <code>README.md</code> file is included, containing detailed documentation of the dataset structure,  data augmentation procedures, and usage instructions.</p> <p>This dataset is intended to develop and evaluate machine learning and deep learning models for Sinhala handwritten text recognition, including applications such as optical character recognition (OCR), image classification, and sequence modeling.</p> <p>For additional details, updates, and related resources, please refer to the GitHub repository: <a href="https://github.com/MaheeshaSewmini/Sinhala-OCR-Dataset">https://github.com/MaheeshaSewmini/Sinhala-OCR-Dataset</a></p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_19878034
institution Zenodo
language eng
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle A Benchmark Dataset of Sinhala Handwritten Text for Optical Character Recognition (OCR) System Development
Rathnayaka, Maheesha Sewmini
Optical Character Recognition
Sinhala
Handwritten text
<p>This dataset contains a collection of images representing Sinhala handwritten characters, words, and paragraphs, covering all characters and connectors. It includes 25,516 scanned images gathered from 381 participants, representing diverse handwriting styles across different demographic groups.</p> <p>This dataset was developed as part of a research project on Sinhala handwritten text recognition.</p> <p>The dataset is annotated with labels at the character, word, and paragraph levels, and includes linked metadata (age, gender, education/occupation category, residence type, and dominant hand) using anonymized participant IDs. Demographic information is recorded separately in an Excel file (<code>MetaData.xlsx</code>) to ensure anonymity.</p> <p>Images were scanned at 300 DPI and split into training, validation, and test sets. Data augmentation was applied only to the training dataset to improve model generalization. The <code>Raw</code> folder contains the original training, validation, and test datasets, while the <code>Augmented</code> folder contains the augmented version of the training dataset.</p> <p>A <code>README.md</code> file is included, containing detailed documentation of the dataset structure,  data augmentation procedures, and usage instructions.</p> <p>This dataset is intended to develop and evaluate machine learning and deep learning models for Sinhala handwritten text recognition, including applications such as optical character recognition (OCR), image classification, and sequence modeling.</p> <p>For additional details, updates, and related resources, please refer to the GitHub repository: <a href="https://github.com/MaheeshaSewmini/Sinhala-OCR-Dataset">https://github.com/MaheeshaSewmini/Sinhala-OCR-Dataset</a></p>
title A Benchmark Dataset of Sinhala Handwritten Text for Optical Character Recognition (OCR) System Development
topic Optical Character Recognition
Sinhala
Handwritten text
url https://doi.org/10.5281/zenodo.19878034