Region-Specific Noise Generation for Untransferable Examples Against Neural Style Transfer (신경망 기반 스타일 변환을 방해할 수 있는 영역별 노이즈 생성 기법)

Sunghwan Park (박성환); Jongseong Kim (김종성); Byunghoon Oh (오병훈); Yohan Hwang (황요한); Junho Hong (홍준호); Jaewoo Lee (이재우)

doi:10.29056/jsav.2025.12.06

Region-Specific Noise Generation for Untransferable Examples Against Neural Style Transfer

Journal of Software Forensics
Abbr : JSF
2025, 21(4), pp.63~77
DOI : 10.29056/jsav.2025.12.06
Publisher : Korea Software Assessment and Valuation Society
Research Area : Engineering > Computer Science
Received : October 20, 2025
Accepted : December 20, 2025
Published : December 26, 2025

Sunghwan Park ¹, Jongseong Kim ¹, Byunghoon Oh ¹, Yohan Hwang ², Junho Hong ³, Jaewoo Lee ¹

¹중앙대학교
²중앙대학교 산업보안연구소
³성신여자대학교

Accredited

ABSTRACT

Neural Style Transfer poses a significant threat to the intellectual property rights of digital artists, as it can extract and replicate unique artistic styles without consent or compensation. Existing protection techniques face a fundamental trade-off: they either degrade image quality for effective protection or offer minimal protection to preserve quality. This paper proposes a novel region-specific noise generation that solves this trade-off by spatially segregating protection objectives. Our method divides an image into two distinct regions and applies different perturbation strategies to each: a small patch region (2% of image area) receives strong perturbations to hijack the attention mechanism, while the background region (98%) is subjected to minimal perturbations to corrupt global feature statistics. The key innovation lies in the spatial separation that fundamentally eliminates gradient interference between different loss objectives. Experimental results demonstrate that our method achieves over 90% reduction in style transfer effectiveness (STDR 0.8995) while maintaining high visual quality (SSIM 0.8624), representing a 38% improvement in protection efficiency compared to existing methods.

KEYWORDS

Neural Style Transfer, Untransferable Examples, Adversarial Attack, Region-specific optimization, Copyright Content Protection

Journal Copyright Policy

No CCL information provided

Citation status

* References for papers published after 2025 are currently being built.

[confproc] Gatys Leon A. / 2016 / Image style transfer using convolutional neural networks / Proceedings of the IEEE conference on computer vision and pattern recognition : 2414~2423

[confproc] Johnson, Justin / 2016 / Perceptual losses for real-time style transfer and super-resolution / European conference on computer vision : 694~711

[confproc] Huang, Xun / 2017 / Arbitrary style transfer in real-time with adaptive instance normalization / Proceedings of the IEEE international conference on computer vision : 1501~1510

[confproc] Park, Dae Young / 2019 / Arbitrary style transfer with style-attentional networks / proceedings of the IEEE/CVF conference on computer vision and pattern recognition : 5880~5888

[confproc] Deng, Yingying / 2022 / Stytr2: Image style transfer with transformers / Proceedings of the IEEE/CVF conference on computer vision and pattern recognition : 11326~11336

[confproc] Zhang, Chiyu / 2024 / S2wat: Image style transfer via hierarchical vision transformer using strips window attention / Proceedings of the AAAI conference on artificial intelligence 38(7) : 7024~7032

[journal] Jafari, Reza / 2023 / Legal Frameworks for Protecting Digital Art and NFTs: Navigating Copyright and Ownership Rights in Virtual Spaces / Legal Studies in Digital Age 2(3) : 25~36

[confproc] Foerster Hanna / 2025 / {LightShed}: Defeating Perturbation-based Image Copyright Protections / 34th USENIX Security Symposium (USENIX Security 25) : 7271~7290

[confproc] Zhang,Dawen / 2024 / Privacy and copyright protection in generative AI: A lifecycle perspective / Proceedings of the IEEE/ACM 3rd International Conference on AI Engineering-Software Engineering for AI : 92~97

[other] Gatys Leon A. / 2016 / Preserving color in neural artistic style transfer

[confproc] Addepalli, Sravanti / 2022 / Scaling adversarial training to large perturbation bounds / European Conference on Computer Vision : 301~316

[confproc] Dewi, V. / 2022 / Legal protection of digital art copyrights on social media / Proceedings of the International Conference on Sustainability in Technological, Environmental, Law, Management, Social and Economic Matters, ICOSTELM : 4~5

[other] NARASIMHAN, Sharan / 2022 / Towards robust and semantically organised latent representations for unsupervised text style transfer

[journal] Luo, Xin / 2023 / Adversarial style discrepancy minimization for unsupervised domain adaptation / Neural Networks 157 : 216~225

[other] Guo Zhongliang / 2024 / Artwork protection against neural style transfer using locally adaptive adversarial color attack

[confproc] Li, Yaxin / 2024 / Neural style protection: Counteracting unauthorized neural style transfer / Proceedings of the IEEE/CVF winter conference on applications of computer vision : 3966~3975

[other] Bala, Aniruddha / 2025 / Dct-shield: A robust frequency domain defense against malicious image editing

[journal] Kang / 2025 / Rect-ViT: Rectified attention via feature attribution can improve the adversarial robustness of Vision Transformers / Neural Networks 190

[confproc] Li, Yanghao / 2017 / Demystifying neural style transfer / Proceedings of the 26th International Joint Conference on Artificial Intelligence : 2230~2236

[confproc] Gardner Steven / 2019 / Constrained multi-objective optimization for automated machine learning / 2019 IEEE International conference on data science and advanced analytics (DSAA) : 364~373

[other] Neel, Seth / 2023 / Privacy issues in large language models: A survey

[journal] Watiktinnakorn, Chawinthorn / 2023 / Blurring the lines: how AI is redefining artistic ownership and copyright / Discover Artificial Intelligence 3(37)

[confproc] Shan, Shawn / 2023 / Glaze: Protecting artists from style mimicry by {Text-to-Image} models / 32nd USENIX Security Symposium (USENIX Security 23) : 2187~2204

[confproc] Li, Jiacheng / 2024 / {MIST}: Defending against membership inference attacks through {Membership-Invariant} subspace training / 33rd USENIX Security Symposium (USENIX Security 24) : 2387~2404

[web] Patrick K. Lin / Can This Data Poisoning Tool Help Artists Protect Their Work from AI Scraping? / Center for Art Law / https://itsartlaw.org/art-law/can-this-data-poisoning-tool-help-artists-protect-their-work-from-ai-scraping/

[other] Zhang, Yunming / 2024 / StyleMark: A Robust Watermarking Method for Art Style Images Against Black-Box Arbitrary Style Transfer

KJCKorea
Journal Central

Journal of Software Forensics 2025 KCI Impact Factor : 0.3

Region-Specific Noise Generation for Untransferable Examples Against Neural Style Transfer

ABSTRACT

KEYWORDS

Journal Copyright Policy

Citation status

* References for papers published after 2025 are currently being built.

Journal of Software Forensics 2025 KCI Impact Factor : 0.3

Region-Specific Noise Generation for Untransferable Examples Against Neural Style Transfer

ABSTRACT

KEYWORDS

Journal Copyright Policy

Statistics

Tools

Issue List

Citation status

KCI Citation Counts (0)

REFERENCES (26) * References for papers published after 2025 are currently being built.

Search PDF

Citation

* References for papers published after 2025 are currently being built.