This is the Github reposipory of the paper "Generalized W-Net: Arbitrary-style Chinese Character Synthesization". Please refer to Scripts/xx.sh for implementation.
Some specific args: encoder: Cv=Conventional convolution basic block (without residual connectionb); Cbb: basic block with residual connection; Cbn: bottleneck Vit@A@B: Vision Transformer block with depth=A and number of heads = B
decoder: in default the decoder will be constructed with the symmetrical architecture; otherwise, follow the encoder;
mixer: Smp: simple mixer Res@A@B: residual mixer with A blocks at B stage
skipTest=True: no full-set testing during the training
Relevant paper can be found:
Generalized W-Net: Arbitrary-style Chinese Character Synthesization https://arxiv.org/abs/2406.06847
W-Net: One-Shot Arbitrary-Style Chinese Character Generation with Deep Neural Networks https://arxiv.org/abs/2406.06122
To cite them:
@inproceedings{jiang2018w,
title={W-net: one-shot arbitrary-style Chinese character generation with deep neural networks},
author={Jiang, Haochuan and Yang, Guanyu and Huang, Kaizhu and Zhang, Rui},
booktitle={Neural Information Processing: 25th International Conference, ICONIP 2018, Siem Reap, Cambodia, December 13--16, 2018, Proceedings, Part V 25},
pages={483--493},
year={2018},
organization={Springer}
}
@inproceedings{jiang2023generalized,
title={Generalized W-Net: Arbitrary-Style Chinese Character Synthesization},
author={Jiang, Haochuan and Yang, Guanyu and Cheng, Fei and Huang, Kaizhu},
booktitle={International Conference on Brain Inspired Cognitive Systems},
pages={190--200},
year={2023},
organization={Springer}
}