DNA: Dual-branch Network with Adaptation for Open-Set Online Handwriting Generation

2025-11-26

AI Summary

Addressing the challenge of joint generalization to unseen characters and writer styles in open-set online handwriting generation (OHG), this paper proposes a dual-branch adaptive network. The style branch models stroke-level dynamics, while the content branch disentangles character structure from texture details. A local-global encoder architecture separately extracts structural and textural representations; adaptive instance normalization and attention mechanisms jointly enable precise style transfer and faithful content reconstruction. Evaluated on multiple Chinese online handwriting datasets, our method significantly outperforms existing approaches, demonstrating superior generation quality and robustness under both unseen-character and unseen-writer settings. To the best of our knowledge, this is the first work to systematically resolve the open-set generalization problem for glyph-based languages in OHG.

Technology Category

• Natural Language Processing: Generation
• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Online handwriting generation (OHG) enhances handwriting recognition models by synthesizing diverse, human-like samples. However, existing OHG methods struggle to generate unseen characters, particularly in glyph-based languages like Chinese, limiting their real-world applicability. In this paper, we introduce our method for OHG, where the writer's style and the characters generated during testing are unseen during training. To tackle this challenge, we propose a Dual-branch Network with Adaptation (DNA), which comprises an adaptive style branch and an adaptive content branch. The style branch learns stroke attributes such as writing direction, spacing, placement, and flow to generate realistic handwriting. Meanwhile, the content branch is designed to generalize effectively to unseen characters by decomposing character content into structural information and texture details, extracted via local and global encoders, respectively. Extensive experiments demonstrate that our DNA model is well-suited for the unseen OHG setting, achieving state-of-the-art performance.

Problem

• Generates unseen characters in online handwriting
• Adapts to new writer styles not seen in training
• Decomposes character content into structure and texture details

Innovation

• Dual-branch network with adaptive style and content branches
• Style branch learns stroke attributes for realistic handwriting
• Content branch decomposes characters into structural and texture details