Variational Grid Setting Network

TL;DR

This paper introduces the Variational Grid Setting Network, a VAE-based architecture capable of generating large missing Chinese characters using minimal training data, advancing font completion technology.

Contribution

The paper presents a novel VAE-based neural network architecture specifically designed for generating large Chinese characters with limited training samples.

Findings

Effective generation of 256x256 pixel characters

High-quality results with few training samples

Applicable to large-scale Chinese font completion

Abstract

We propose a new neural network architecture for automatic generation of missing characters in a Chinese font set. We call the neural network architecture the Variational Grid Setting Network which is based on the variational autoencoder (VAE) with some tweaks. The neural network model is able to generate missing characters relatively large in size ($256 \times 256$ pixels). Moreover, we show that one can use very few samples for training data set, and get a satisfied result.