Skip to content
master
Go to file
Code

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
Jan 29, 2018

README.md

HCCG-CycleGAN

Generating Handwritten Chinese Characters using CycleGAN
Bo Chang*, Qiong Zhang*, Shenyi Pan, Lili Meng
WACV 2018.

Abstract

Handwriting of Chinese has long been an important skill in East Asia. However, automatic generation of handwritten Chinese characters poses a great challenge due to the large number of characters. Various machine learning techniques have been used to recognize Chinese characters, but few works have studied the handwritten Chinese character generation problem, especially with unpaired training data. In this work, we formulate the Chinese handwritten character generation as a problem that learns a mapping from an existing printed font to a personalized handwritten style. We further propose DenseNet CycleGAN to generate Chinese handwritten characters. Our method is applied not only to commonly used Chinese characters but also to calligraphy work with aesthetic values. Furthermore, we propose content accuracy and style discrepancy as the evaluation metrics to assess the quality of the handwritten characters generated. We then use our proposed metrics to evaluate the generated characters from CASIA dataset as well as our newly introduced Lanting calligraphy dataset.

Requirement

Python 2.7

Preprocess handwritting database

The gnt files in CASIA HWDB1.1 database can be downloaded here.

python preprocess/preprocess_hw.py fonts/1252-c.gnt fonts/simhei.ttf 0.1 0.1 

Both the gnt file for the handwritten data and the ttf/otf file for the source font need to be under the fonts folder. The last two number represents for the split ratio of the training and test data.

DenseCycleGAN

Usage

  • Training
cd DenseCycleGAN
DATA_ROOT=./datasets/1252-c-0.1-0.1 name=1252-c-dense5 which_model_netG=densenet_5blocks th train.lua
  • Testing
DATA_ROOT=./datasets/1252-c-0.1-0.1 name=1252-c-dense5 phase=test th test.lua

Supported which_model_netG in training

  • resnet_6blocks
  • resnet_9blocks
  • densenet_5blocks
  • densenet_6blocks
  • densenet_7blocks
  • densenet_8blocks

BibTeX

@inproceedings{chang2018generating, 
  title={Generating Handwritten Chinese Characters Using CycleGAN}, 
  author={Chang, Bo and Zhang, Qiong and Pan, Shenyi and Meng, Lili}, 
  booktitle={IEEE Winter Conference on Applications of Computer Vision}, 
  year={2018} 
}

About

Handwritten Chinese Characters Generation

Topics

Resources

Releases

No releases published

Packages

No packages published