VNOnDB Online to Offline dataset converter

Introduction

HANDS-VNOnDB2018 (VNOnDB2018 in short) is used for ICFHR2018 Competition on Vietnamese Online Handwritten Text Recognition using VNOnDB. It provides 1,146 Vietnamese paragraphs of handwritten text composed of 7,296 lines, more than 480,000 strokes and more that 380,000 characters written by 200 Vietnamese.

For more information about this dataset, please go to the main page

Examples

With dpi=300, below are some examples from database after converted

Level	line_width=1	line_width=2
word
line
paragraph

Prerequisite

Install libraries

• pip install -r requirements.txt

You also have dataset and datasplit downloaded and extracted. Your directory might as follow:

.git/
.gitignore
data/InkData_line/
data/InkData_paragraph/
data/InkData_word/
data/VNOnDB_ICFHR2018_dataSplit/
images/
LICENSE
convert.py
convert.ipynb
README.md
requirements.txt

Run

Note: This is a very time consuming process!!

Help

• python convert.py --help for help

Quick run!

• To convert line
  • python convert.py line
• To convert word
  • python convert.py word
• To convert paragraph
  • python convert.py paragraph

Note: Results *.csv files and image folders will be stored in ./data

Detail

python convert.py --line_width <line_width> --dpi <dpi> --label_only

• (Optional) <line_width> is line stroke width, default is 2
• (Optional) <dpi>, default is 300
• (Optional) --label_only, if specified, this will only generate *.csv. This is useful in case you already have images and just need the label files.