Watermarks vs. Perturbations for Preventing AI-based Style Editing
GenAI Watermarking Workshop at ICLR · 2025
Compares traditional watermarking and adversarial protection methods for preventing unauthorized style editing with Stable Diffusion.
Research Output
Publications
Browse publications, preprints, and projects. Use the filters to focus on conference, workshop, or preprint entries.
Publications
Is Perturbation-Based Image Protection Disruptive to Image Editing?
IEEE International Conference on Image Processing · 2025
Perturbation-based image protection methods often fail against diffusion-based editing and can even unintentionally strengthen prompt-guided edits instead of preventing them.
Exploring Saliency Bias in Manipulation Detection
IEEE International Conference on Image Processing · 2024
Studies how semantic and visual saliency influence image manipulation detection and motivates more semantic-aware forensic analysis.
Preprints
MicroFakes: A Dataset for Small but Semantically Relevant Image Manipulations
arXiv · 2026
Introduces a dataset for small but semantically relevant image manipulations.
ReVA: A New Scene-Centric Dataset for Remote Sensing Video Question Answering
arXiv · 2026
Introduces a large-scale dataset for remote sensing video question answering, designed to assess spatiotemporal, scene-centric, and reasoning-oriented capabilities of MLLMs.
A Survey of Origin Protection Methods against Diffusion-based Image Editing
arXiv · 2026
Provides a systematic and structured overview of protection mechanisms against diffusion-based image manipulation, organized along three dimensions: threat model, victim image type, and protection paradigm.
StyleProtect: Safeguarding Artistic Identity in Fine-tuned Diffusion Models
arXiv · 2025
Introduces a lightweight protection strategy for defending artistic style against imitation through fine-tuned diffusion models.
Projects
Projects will appear here as they are added.