Research map / Image generation / Editing and style transfer
Few-shot Font Generation with Localized Style Representations and Factorization
Song Park, Sanghyuk Chun, Junbum Cha, Bado Lee, Hyunjung Shim
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 localized component-wise style representations instead of universal style encoding (ablation) | LF-Font · Chinese unseen chars | Acc (S): 33.6 -> 72.8 | Improved |
| Add component-wise encoder E_s,u to universal style encoding (ablation) | LF-Font style encoder · Chinese unseen chars | Acc (S): 33.6 -> 52.8 | Improved |
| Add content encoder E_c (ablation) | LF-Font · Chinese unseen chars | Acc (Hmean): 21.7% -> 83.2% | Improved |
| Add consistency loss L_consist while omitting L_cls (ablation) | LF-Font · Chinese unseen chars | Acc (Hmean): 56.2 -> 61.6 | Improved |
| Add component classification loss L_cls while omitting L_consist (ablation) | LF-Font · Chinese unseen chars | Acc (Hmean): 56.2 -> 81.1 | Improved |
| Add component classification loss L_cls to the consistency-loss model (ablation) | LF-Font · Chinese unseen chars | Acc (Hmean): 61.6 -> 83.2 | Improved |
| Set factor dimension k to 4 instead of 8 (ablation) | LF-Font · Chinese unseen chars | Acc (Hmean): 83.2 -> 82.3 | Hurt |
| Set factor dimension k to 6 instead of 8 (ablation) | LF-Font · Chinese unseen chars | Acc (Hmean): 83.2 -> 83.0 | Hurt |
| Set factor dimension k to 10 instead of 8 (ablation) | LF-Font · Chinese unseen chars | Acc (Hmean): 83.2 -> 82.4 | Hurt |
| Use LF-Font rather than the strongest baseline FUNIT | few-shot font generation · Chinese seen chars | Acc (Hmean): 55.7 -> 84.8 | Improved |
| Use LF-Font rather than the strongest baseline FUNIT | few-shot font generation · Chinese unseen chars | Acc (Hmean): 54.5 -> 83.2 | Improved |
| Use LF-Font rather than AGIS-Net | few-shot font generation · Korean | Acc (Hmean): 7.5 -> 58.5 | Improved |
Closest papers by meaning
- VQ-FONT: Few-Shot Font Generation with Structure-Aware Enhancement and Quantization (AAAI 2024 · similarity 0.91)
- IF-Font: Ideographic Description Sequence-Following Font Generation (NeurIPS 2024 · similarity 0.90)
- DeepCalliFont: Few-Shot Chinese Calligraphy Font Synthesis by Integrating Dual-Modality Generative Models (AAAI 2024 · similarity 0.89)
- FontDiffuser: One-Shot Font Generation via Denoising Diffusion with Multi-Scale Content Aggregation and Style Contrastive Learning (AAAI 2024 · similarity 0.89)
- Generate Like Experts: Multi-Stage Font Generation by Incorporating Font Transfer Process into Diffusion Models (CVPR 2024 · similarity 0.88)
- Decoupling Layout from Glyph in Online Chinese Handwriting Generation (ICLR 2025 · similarity 0.88)
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.
