HANDS-VNOnDB: TUAT Nakagawa Lab. Vietnamese Online Handwriting Database
Authors/Creators
Contributors
Hosting institution:
Description
HANDS-VNOnDB: Vietnamese Online Handwriting Database
Dataset Description
HANDS-VNOnDB (VNOnDB) is a database of unconstrained Vietnamese online handwritten text designed and collected by the Nakagawa Laboratory at Tokyo University of Agriculture and Technology (TUAT), Japan.
The database was developed to support research on online handwriting recognition, particularly the challenges associated with Vietnamese script, which contains numerous diacritical marks and frequently exhibits delayed strokes.
VNOnDB contains handwriting collected from 200 Vietnamese writers, comprising:
-
1,146 handwritten paragraphs
-
7,296 handwritten text lines
-
More than 480,000 pen strokes
-
More than 380,000 characters
Participants were asked to naturally reproduce Vietnamese text selected from a corpus derived from VieTreeBank, which includes Vietnamese characters and special symbols found in newspaper texts. Handwriting was collected using Fujitsu Tablet PCs (FMVT8170) equipped with stylus pens, at a sampling rate of 120 Hz.
Data Format and Annotations
Handwriting data are stored in Ink Markup Language (InkML) format, preserving the ordered pen trajectories as sequences of coordinate points.
The database provides handwritten patterns and corresponding ground-truth transcriptions at different levels, including paragraphs, lines, and words.
InkML files also contain metadata, such as writer identifiers, ground-truth references, and available writer attributes, including age, gender, dominant hand, writing hand, occupation, and native language.
These characteristics make VNOnDB suitable for research on online handwritten text recognition, sequence modeling, stroke-order analysis, delayed-stroke processing, and writer-independent recognition.
Dataset Releases
HANDS-VNOnDB (2016)
The original release provides paragraph-level and line-level handwriting data with corresponding ground-truth transcriptions and predefined training, validation, and test splits.
HANDS-VNOnDB2018 (ICFHR 2018)
The ICFHR 2018 competition release provides handwriting data organized at paragraph, line, and word levels, together with the training, validation, and test split definitions used in the ICFHR 2018 Competition on Vietnamese Online Handwritten Text Recognition.
Dataset Access and Usage Agreement
VNOnDB is available upon request for non-commercial research purposes only.
To obtain access to the database, prospective users must review, sign, and return the Dataset Usage Agreement provided by the dataset administrators.
Dataset access will be granted only after the signed agreement has been received and the request has been approved.
All users must comply with the terms and conditions specified in the signed Dataset Usage Agreement. Commercial use, including commercial model development and commercial recognition services, is not covered by the standard research-use permission and requires separate authorization from the dataset owners.
Researchers interested in accessing VNOnDB should contact the dataset administrators to request the Dataset Usage Agreement and further instructions.
Citation
Our paper provides a detailed description of the database construction, collection methodology, statistical analysis, characteristics of Vietnamese handwriting, and baseline recognition experiments. Researchers using VNOnDB in publications, theses, technical reports, or other scholarly outputs are required to acknowledge the database and cite the following publication:
Nguyen, H. T., Nguyen, C. T., Bao, P. T., & Nakagawa, M. (2018). A database of unconstrained Vietnamese online handwriting and recognition experiments by recurrent neural networks. Pattern Recognition, 78, 291–306.
Dataset Contact
Nakagawa Laboratory, Tokyo University of Agriculture and Technology (TUAT), Japan
Dataset contact author: Hung Tuan Nguyen (ntuanhung@gmail.com)
Please contact the dataset administrators for access requests, usage agreements, and commercial licensing inquiries.
Files
Additional details
Related works
- Is described by
- Journal article: 10.1016/j.patcog.2018.01.013 (DOI)
Dates
- Available
-
2016-11-25First online available
- Updated
-
2018-02-21Supervised labeling for Word-, Line-, and Paragraph-level
- Copyrighted
-
2026-11-02Update the "conditions of use" and application form