Research map / Image generation / Editing and style transfer
UniCalli: A Unified Diffusion Framework for Column-Level Generation and Recognition of Chinese Calligraphy
Tianshuo Xu, Kai Wang, ZhiFei Chen, Leyi Wu, Tianshui Wen, Fei Chao, Ying-Cong Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use UniCalli instead of Doubao for generation | UniCalli · Calligraphy generation user study | Style Fidelity: 2.413 -> 4.267 | Improved |
| use UniCalli instead of FontDiffuser for generation | UniCalli · Calligraphy generation user study | Glyph Accuracy: 4.950 -> 4.827 | No clear effect |
| use UniCalli instead of Doubao for generation | UniCalli · Calligraphy generation user study | Overall Preference: 3.933 -> 4.560 | Improved |
| use UniCalli instead of CalliReader for recognition | UniCalli · Character recognition total | character accuracy: 0.533 -> 0.540 | Improved |
| use UniCalli instead of Ernie-4.5 for Cursive/Cao recognition | UniCalli · Character recognition Cursive/Cao | character accuracy: 0.255 -> 0.109 | Hurt |
| use UniCalli instead of CalliReader for Running/Xing recognition | UniCalli · Character recognition Running/Xing | character accuracy: 0.658 -> 0.528 | Hurt |
| use UniCalli instead of OracleNet for oracle bone recognition | UniCalli · Oracle bone script recognition | recognition accuracy (%): 73.9 -> 62.5 | Hurt |
| use UniCalli instead of GOT-OCR2.0 on out-of-domain CalliBench | UniCalli · CalliBench Easy | F1: 0.593 -> 0.498 | Hurt |
| use UniCalli instead of GPT-4o on CalliBench | UniCalli · CalliBench Easy | F1: 0.400 -> 0.498 | Improved |
| set conditional dropout rate to 0.05 rather than higher or lower (ablation) | UniCalli · calligraphy generation | style and glyph quality | Improved |
| add Duplicate RoPE with Modulated Embedding, joint training and conditional dropout to FLUX baseline (ablation) | UniCalli · calligraphy generation | L1, SSIM, LPIPS, FID | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- FontDiffuser — FontDiffuser: One-Shot Font Generation via Denoising Diffusion with Multi-Scale Content Aggregation and Style Contrastive Learning (AAAI 2024) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Towards Automated Chinese Ancient Character Restoration: A Diffusion-Based Method with a New Dataset (AAAI 2024 · similarity 0.89)
- Decoupling Layout from Glyph in Online Chinese Handwriting Generation (ICLR 2025 · similarity 0.88)
- DeepCalliFont: Few-Shot Chinese Calligraphy Font Synthesis by Integrating Dual-Modality Generative Models (AAAI 2024 · similarity 0.87)
- A Multi-Style Chinese Characters Writing Intelligent Tool Based on Small-scale Training Data (AAAI 2025 · similarity 0.86)
- OBI-Bench: Can LMMs Aid in Study of Ancient Script on Oracle Bones? (ICLR 2025 · similarity 0.84)
- Component-Level Segmentation for Oracle Bone Inscription Decipherment (AAAI 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
