Research map / Image generation / Editing and style transfer
Multiple Heads Are Better Than One: Few-Shot Font Generation With Multiple Localized Experts
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 6 localized experts instead of 1 (ablation) | MX-Font · Chinese to Chinese (in-domain) | Acc (B) %: 71.4 -> 78.7 | Improved |
| use 6 localized experts instead of 1 (ablation) | MX-Font · Chinese to Chinese (in-domain) | LPIPS: 0.133 -> 0.120 | Improved |
| increase experts from 6 to 8 (ablation) | MX-Font · Chinese to Chinese (in-domain) | Acc (B) %: 78.7 -> 75.2 | Hurt |
| increase experts from 2 to 6 (ablation) | MX-Font · Chinese to Chinese (in-domain) | Acc (B) %: 78.5 -> 78.7 | No clear effect |
| use component classifier instead of character classifier as weak supervision (ablation) | MX-Font · Chinese to Chinese (in-domain) | Acc (B) %: 0.04 -> 78.7 | Improved |
| add independence loss (ablation) | MX-Font · Korean handwriting (loss ablation) | Acc (B) %: 50.0 -> 56.8 | Improved |
| add maximizing entropy term of content-style adversarial loss (ablation) | MX-Font · Korean handwriting (loss ablation) | Acc (B) %: 49.4 -> 50.0 | No clear effect |
| add content-style adversarial loss (ablation) | MX-Font · Korean handwriting (loss ablation) | Acc (B) %: 24.7 -> 49.4 | Improved |
| MX-Font vs LF-Font in cross-lingual generation | MX-Font · Chinese to Korean | Acc (B) %: 12.8 -> 50.0 | Improved |
| MX-Font vs LF-Font on Chinese FID | MX-Font · Chinese to Chinese (in-domain) | FID (H): 14.8 -> 21.8 | Hurt |
| MX-Font vs LF-Font user preference on Chinese | MX-Font · Chinese to Chinese (in-domain) | User preference (B) %: 15.6 -> 35.2 | Improved |
Closest papers by meaning
- Few-shot Font Generation with Localized Style Representations and Factorization (AAAI 2021 · similarity 0.94)
- IF-Font: Ideographic Description Sequence-Following Font Generation (NeurIPS 2024 · similarity 0.90)
- Generate Like Experts: Multi-Stage Font Generation by Incorporating Font Transfer Process into Diffusion Models (CVPR 2024 · similarity 0.90)
- VQ-FONT: Few-Shot Font Generation with Structure-Aware Enhancement and Quantization (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 0.88)
- DeepCalliFont: Few-Shot Chinese Calligraphy Font Synthesis by Integrating Dual-Modality Generative Models (AAAI 2024 · similarity 0.87)
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.
