Research map / Image generation / Editing and style transfer
Towards Ultra-Resolution Neural Style Transfer via Thumbnail Instance Normalization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 patch-wise processing with TIN instead of full-image processing | Existing IN-based style transfer methods · 10000² images | GPU memory (GB): 30.25 -> 1.94 | Helped |
| Use patch-wise TIN rather than independently normalized patches (ablation) | Patch-wise style transfer · Ultra-high-resolution images | style consistency | Helped |
| Replace thumbnail statistics with random normalization statistics (ablation) | Patch-wise style transfer · Ultra-high-resolution images | style fidelity | Hurt |
| Increase thumbnail size to larger than 1024×1024 (ablation) | TIN · 8192×8192 input image | normalization statistics similarity to IN | Helped |
| Add stroke perceptual loss to the style-transfer objective (ablation) | Existing style transfer models · Ultra-high-resolution style transfer | brush-stroke size | Helped |
| Add stroke perceptual loss to the style-transfer objective (ablation) | Existing style transfer models · Ultra-high-resolution style transfer | stylization quality | Helped |
| Apply URST patch-wise processing | Johnson · 10000² image | GPU memory (GB): - -> 2.12 | Helped |
| Apply URST patch-wise processing | AdaIN · 10000² image | GPU memory (GB): - -> 2.17 | Helped |
Closest papers by meaning
- MicroAST: Towards Super-fast Ultra-Resolution Arbitrary Style Transfer (AAAI 2023 · similarity 0.90)
- Collaborative Distillation for Ultra-Resolution Universal Style Transfer (CVPR 2020 · similarity 0.88)
- Universal Style Transfer via Feature Transforms (NeurIPS 2017 · similarity 0.87)
- Dynamic Instance Normalization for Arbitrary Style Transfer (AAAI 2020 · similarity 0.87)
- Learning Linear Transformations for Fast Image and Video Style Transfer (CVPR 2019 · similarity 0.87)
- Style-Aware Normalized Loss for Improving Arbitrary Style Transfer (CVPR 2021 · 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.