Asset Protection

Generative models make it easy to copy an artist’s style or edit a personal photo without consent. We study how to protect visual assets from unauthorized AI-based editing and mimicry, from imperceptible adversarial perturbations to watermarking, and we critically evaluate whether these protections actually hold up against the editing pipelines they are meant to disrupt.

Publications

StyleProtect: Safeguarding Artistic Identity in Finetuned Diffusion Models

Qiuyu Tang, Joshua Krinsky, Aparna Bharati

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  ·  03 Jun 2026  ·  arxiv:2509.13711

Project Page

Code

Is Perturbation-Based Image Protection Disruptive to Image Editing?

Qiuyu Tang, Bonor Ayambem, Mooi Choo Chuah, Aparna Bharati

2025 IEEE International Conference on Image Processing (ICIP)  ·  14 Sep 2025  ·  doi:10.1109/icip55913.2025.11084494

Code

Supplement

Watermarks vs. Perturbations for Preventing AI-Based Style Editing

Qiuyu Tang, Aparna Bharati

The 1st Workshop on GenAI Watermarking (WMARK) at ICLR 2025  ·  24 Apr 2025

OpenReview PDF