GC-Font: Few-Shot Font Generation via Global Contextual Feature Modelling


Weiran Chen (Soochow University), Guiqian Zhu (Soochow University), Ying Li (Soochow University), Yi Ji (Soochow University), Chunping Liu (Soochow University)
The 35th British Machine Vision Conference
PDFPosterVideo (Right click to download)Supplementary

Abstract

Few-shot font generation aims to create new fonts with a limited number of glyph references. It can be used to greatly reduce the labour cost of manual font design. However, due to the variety and complexity of font styles, the results generated by existing methods often suffer from visible defects, such as stroke errors, blurriness or distorted shapes. To address these issues, we propose GC-Font, a novel framework which integrates a Global Contextual Feature Modelling (GCFM) module. Specifically, this module is inserted between the content encoder and decoder, where it fuses convolution and attention mechanisms to process intermediate feature maps and injects enhanced global contextual features into the decoder. Moreover, we apply adaptive convolutions to the low-level feature maps from the content encoder to strengthen contextual correlations. In addition, a skeleton consistency loss and an edge consistency loss are also designed to improve geometric alignment. Extensive experiments reveal that our GC-Font outperforms the state-of-the-art methods in both qualitative and quantitative evaluations, demonstrating its effectiveness on diverse font styles. Our source code can be found at https://github.com/wrchen2001/GC-Font.

Citation

@inproceedings{Chen_2025_BMVC,
author    = {Weiran Chen and Guiqian Zhu and Ying Li and Yi Ji and Chunping Liu},
title     = {GC-Font: Few-Shot Font Generation via Global Contextual Feature Modelling},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_38/paper.pdf}
}


Copyright © 2025 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection