Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generalized W-Net: Arbitrary-style Chinese Character Synthesization

Haochuan Jiang, Guanyu Yang, Fei Cheng, Kaizhu Huang

TL;DR

The Generalized W-Net is proposed, a novel class of W-shaped architectures that addresses synthesizing Chinese characters in any desired style, even with limited examples, by incorporating Adaptive Instance Normalization and introducing multi-content.

Abstract

Synthesizing Chinese characters with consistent style using few stylized examples is challenging. Existing models struggle to generate arbitrary style characters with limited examples. In this paper, we propose the Generalized W-Net, a novel class of W-shaped architectures that addresses this. By incorporating Adaptive Instance Normalization and introducing multi-content, our approach can synthesize Chinese characters in any desired style, even with limited examples. It handles seen and unseen styles during training and can generate new character contents. Experimental results demonstrate the effectiveness of our approach.

Generalized W-Net: Arbitrary-style Chinese Character Synthesization

TL;DR

The Generalized W-Net is proposed, a novel class of W-shaped architectures that addresses synthesizing Chinese characters in any desired style, even with limited examples, by incorporating Adaptive Instance Normalization and introducing multi-content.

Abstract

Synthesizing Chinese characters with consistent style using few stylized examples is challenging. Existing models struggle to generate arbitrary style characters with limited examples. In this paper, we propose the Generalized W-Net, a novel class of W-shaped architectures that addresses this. By incorporating Adaptive Instance Normalization and introducing multi-content, our approach can synthesize Chinese characters in any desired style, even with limited examples. It handles seen and unseen styles during training and can generate new character contents. Experimental results demonstrate the effectiveness of our approach.
Paper Structure (19 sections, 3 equations, 7 figures, 1 table)

This paper contains 19 sections, 3 equations, 7 figures, 1 table.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. The Generalized W-Net Architecture
  4. Encoders
  5. Feature Mixer
  6. AdaIN in the Feature Mixer
  7. AdaIN on the Extracted Content Feature
  8. AdaIN in the Dense / Residual Blocks
  9. Other Corporations
  10. Character Decoder
  11. Training Strategy and Optimization Losses
  12. Training Strategy
  13. Adversarial Loss
  14. Categorical Loss of the Auxiliary Classifier
  15. Constant Losses of the Encoders
  16. ...and 4 more sections

Figures (7)

  • Figure 1: Examples of Chinese characters and calligraphic work.
  • Figure 2: Structure of W-shaped architectures for the character generation task.
  • Figure 3: Model Architecture: Generalized W-Net
  • Figure 4: Examples of Chinese characters and calligraphic work.
  • Figure 5: The input few brush-written examples of actual Chinese characters.
  • ...and 2 more figures