Research map / Image generation / Editing and style transfer
ArtFlow: Unbiased Image Style Transfer via Reversible Neural Flows
Jie An, Siyu Huang, Yibing Song, Dejing Dou, Wei Liu, Jiebo Luo
Check 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 |
|---|---|---|---|
| Replace the VGG19 encoder-decoder with reversible PFN while using AdaIN transfer | AdaIN style transfer · Repeated style transfer (20 rounds) | content leak | Improved |
| Replace the VGG19 encoder-decoder with reversible PFN while using WCT transfer | WCT style transfer · Repeated style transfer (20 rounds) | content leak | Improved |
| Use PFN with AdaIN rather than vanilla AdaIN | AdaIN · Style transfer (512×512) | Gram Loss: 0.00127 -> 0.00098 | Improved |
| Use PFN with WCT rather than vanilla WCT | WCT · Style transfer (512×512) | Gram Loss: 0.00074 -> 0.00078 | No clear effect |
| Use PFN with AdaIN rather than vanilla AdaIN | AdaIN · Style transfer (512×512) | SSIM: 0.29 -> 0.45 | Improved |
| Use PFN with WCT rather than vanilla WCT | WCT · Style transfer (512×512) | SSIM: 0.27 -> 0.47 | Improved |
| Use PFN with AdaIN rather than vanilla AdaIN | AdaIN · Style transfer (512×512) | Time (seconds): 0.064 -> 0.408 | Hurt |
| Use PFN with WCT rather than vanilla WCT | WCT · Style transfer (512×512) | Time (seconds): 0.997 -> 0.428 | Improved |
| Replace PFN (Flow8-Block2) with Flow4-Block4 (ablation) | ArtFlow · Style transfer (512×512) | SSIM: 0.45 -> 0.401 | Hurt |
| Replace PFN (Flow8-Block2) with Flow8-Block1 (ablation) | ArtFlow · Style transfer (512×512) | Gram Loss: 0.00098 -> 0.00139 | Hurt |
| Replace PFN (Flow8-Block2) with Flow16-Block1 (ablation) | ArtFlow · Style transfer (512×512) | Gram Loss: 0.00098 -> 0.00137 | Hurt |
| Replace PFN (Flow8-Block2) with Flow4-Block4 (ablation) | ArtFlow · Style transfer (512×512) | Gram Loss: 0.00098 -> 0.00130 | Hurt |
Later papers that compared with this method
9 later accepted papers report ArtFlow as a baseline in their results tables; 8 of them report a better number on most of the results they share with it.
- An Efficient and Harmonized Framework for Balanced Cross-Domain Feature Integration (AAAI 2026 · better on 4 of 4 shared results)
- StyleFM: Frequency Manipulation Empowered by Recursive Attention on Diffusion Models for Arbitrary Style Transfer (AAAI 2026 · better on 1 of 1 shared result)
- SigStyle: Signature Style Transfer via Personalized Text-to-Image Models (AAAI 2025 · better on 3 of 4 shared results)
- Towards compact reversible image representations for neural style transfer (ECCV 2024 · better on 2 of 2 shared results)
- QuantArt: Quantizing Image Style Transfer Towards High Visual Fidelity (CVPR 2023 · better on 0 of 2 shared results)
- General Image-to-Image Translation with One-Shot Image Guidance (ICCV 2023 · better on 2 of 2 shared results)
- Inversion-Based Style Transfer With Diffusion Models (CVPR 2023 · better on 1 of 1 shared result)
- Industrial Style Transfer With Large-Scale Geometric Warping and Content Preservation (CVPR 2022 · better on 1 of 1 shared result)
1 more not listed.
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 compact reversible image representations for neural style transfer (ECCV 2024 · similarity 0.91)
- A Style-Aware Content Loss for Real-time HD Style Transfer (ECCV 2018 · similarity 0.89)
- Universal Style Transfer via Feature Transforms (NeurIPS 2017 · similarity 0.88)
- Rethinking and Improving the Robustness of Image Style Transfer (CVPR 2021 · similarity 0.88)
- CLeaR: A Unified Framework for Resolving the Leakage–Degradation Dilemma in Style Transfer (NeurIPS 2026 · similarity 0.88)
- Learning Linear Transformations for Fast Image and Video Style Transfer (CVPR 2019 · 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.
