Awesome
π₯ Disentangling Writer and Character Styles for Handwriting Generation
<p align='center'> <b> <a href="https://arxiv.org/abs/2303.14736">ArXiv</a> | <a href="https://github.com/dailenson/SDT/blob/master/static/Poster_SDT.pdf">Poster</a> | <a href="https://youtu.be/mKbYLEwa4dI">Video</a> | <a href="https://cvpr2023.thecvf.com/virtual/2023/poster/20954">Project</a> </b> </p>π’ Introduction
- The proposed style-disentangled Transformer (SDT) generates online handwritings with conditional content and style.
- Existing RNN-based methods mainly focus on capturing a personβs overall writing style, neglecting subtle style inconsistencies between characters written by the same person. In light of this, SDT disentangles the writer-wise and character-wise style representations from individual handwriting samples for enhancing imitation performance.
- We extend SDT and introduce an offline-to-offline framework for improving the generation quality of offline Chinese handwritings.
π News
- [2024/11/26] πππ Release of the implementations of Content Score and Style Score.
- [2024/07/01] πππ A new state-of-the-art method for handwritten text generation, named One-DM, is accepted by ECCV 2024.
- [2024/01/07] Add a tutorial and code for synthesizing handwriting with user-customized styles, more information can be found here.
- [2023/12/15] πππ This work is reported by a top bilibili video blogger with 2.7 million followers and received nearly one million views.
- [2023/10/10] The author is invited to give a talk (in Chinese) by CSIG (China Society of Image and Graphics).
- [2023/06/14] This work is reported by Synced (ζΊε¨δΉεΏ).
- [2023/04/12] Initial release of the datasets, pre-trained models, training and testing codes.
- [2023/02/28] πππ Our SDT is accepted by CVPR 2023.
πΊ Handwriting generation results
-
Online Chinese handwriting generation
-
Applications to various scripts
-
Extension on offline Chinese handwriting generation
π¨ Requirements
conda create -n sdt python=3.8 -y
conda activate sdt
# install all dependencies
conda env create -f environment.yml
π Folder Structure
SDT/
β
βββ train.py - main script to start training
βββ test.py - generate characters via trained model
βββ evaluate.py - evaluation of generated samples
β
βββ configs/*.yml - holds configuration for training
βββ parse_config.py - class to handle config file
β
βββ data_loader/ - anything about data loading goes here
β βββ loader.py
β
βββ model_zoo/ - pre-trained content encoder model
β
βββ data/ - default directory for storing experimental datasets
β
βββ model/ - networks, models and losses
β βββ encoder.py
β βββ gmm.py
β βββ loss.py
β βββ model.py
β βββ transformer.py
β
βββ saved/
β βββ models/ - trained models are saved here
β βββ tborad/ - tensorboard visualization
β βββ samples/ - visualization samples in the training process
β
βββ trainer/ - trainers
β βββ trainer.py
β
βββ utils/ - small utility functions
βββ util.py
βββ logger.py - set log dir for tensorboard and logging output
πΏ Datasets
We provide Chinese, Japanese and English datasets in Google Drive | Baidu Netdisk PW:xu9u. Please download these datasets, uzip them and move the extracted files to /data.
π Pre-trained model
Model | Google Drive | Baidu Netdisk |
---|---|---|
Well-trained SDT | Google Drive | Baidu Netdisk |
Content encoder | Google Drive | Baidu Netdisk |
Content Score | Google Drive | Baidu Netdisk |
Style Score | Google Drive | Baidu Netdisk |
Note: Please download these weights, and move them to /model_zoo.
π Training & Test
Training
- To train the SDT on the Chinese dataset, run this command:
python train.py --cfg configs/CHINESE_CASIA.yml --log Chinese_log
- To train the SDT on the Japanese dataset, run this command:
python train.py --cfg configs/Japanese_TUATHANDS.yml --log Japanese_log
- To train the SDT on the English dataset, run this command:
python train.py --cfg configs/English_CASIA.yml --log English_log
Qualitative Test
- To generate online Chinese handwritings with our SDT, run this command:
python test.py --pretrained_model checkpoint_path --store_type online --sample_size 500 --dir Generated/Chinese
- To generate offline Chinese handwriting images with our SDT, run this command:
python test.py --pretrained_model checkpoint_path --store_type offline --sample_size 500 --dir Generated_img/Chinese
- To generate online Japanese handwritings with our SDT, run this command:
python test.py --pretrained_model checkpoint_path --store_type online --sample_size 500 --dir Generated/Japanese
- To generate offline Japanese handwriting images with our SDT, run this command:
python test.py --pretrained_model checkpoint_path --store_type offline --sample_size 500 --dir Generated_img/Japanese
- To generate online English handwritings with our SDT, run this command:
python test.py --pretrained_model checkpoint_path --store_type online --sample_size 500 --dir Generated/English
- To generate offline English handwriting images with our SDT, run this command:
python test.py --pretrained_model checkpoint_path --store_type offline --sample_size 500 --dir Generated_img/English
Quantitative Evaluation
- To evaluate the generated handwritings, you need to set
data_path
to the path of the generated handwritings (e.g., Generated/Chinese), and run this command:
python evaluate.py --data_path Generated/Chinese --metric DTW
- To calculate the Content Score of generated handwritings, you need to set
data_path
to the path of the generated handwritings (e.g., Generated/Chinese), and run this command:
python evaluate.py --data_path Generated/Chinese --metric Content_score --pretrained_model model_zoo/chinese_content_iter30k_acc95.pth
- To calculate the Style Score of generated handwritings, you need to set
data_path
to the path of the generated handwriting images (e.g., Generated_img/Chinese), and run this command:
python evaluate.py --data_path Generated_img/Chinese --metric Style_score --pretrained_model models_zoo/chinese_style_iter60k_acc999.pth
π° Practical Application
We are delighted to discover that P0etry-rain has proposed a pipeline that involves initially converting the generated results by our SDT to TTF format, followed by the development of software to enable flexible adjustments in spacing between paragraphs, lines, and characters. Below, we present TTF files, software interface and the printed results. More details can be seen in #78.
-
TTF File
-
Software Interface
-
Printed Results
β€οΈ Citation
If you find our work inspiring or use our codebase in your research, please cite our work:
@inproceedings{dai2023disentangling,
title={Disentangling Writer and Character Styles for Handwriting Generation},
author={Dai, Gang and Zhang, Yifan and Wang, Qingfeng and Du, Qing and Yu, Zhuliang and Liu, Zhuoman and Huang, Shuangping},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,
pages={5977--5986},
year={2023}
}