← Search

Lingyun Zhang

2 accepted papers

2026

IdentityGuard: Context-Aware Restriction and Provenance for Personalized Synthesis

ICASSP 2026poster

The nature of personalized text-to-image models poses a unique safety challenge that generic context-blind methods are ill-equipped to handle. Such global filters create a dilemma: to prevent misuse, they are forced to damage the model's broader utility by erasing concepts entirely, causing unaccept…

Cited by 0SourcePDFScholar
2025

Concept Replacer: Replacing Sensitive Concepts in Diffusion Models via Precision Localization

CVPR 2025poster

As large-scale diffusion models continue to advance, they excel at producing high-quality images but often generate unwanted content, such as sexually explicit or violent content. Existing methods for concept removal generally guide the image generation process but can unintentionally modify unrelat…