A perturbation computed from early attention and value matrices can make LLaVA, Instruct-BLIP, and BLIP2-T5 fail to detect objects inside a specified image region while keeping the rest of the image usable.
Effective prompt extraction from language models,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.IV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
VIP: Visual Information Protection through Adversarial Attacks on Vision-Language Models
A perturbation computed from early attention and value matrices can make LLaVA, Instruct-BLIP, and BLIP2-T5 fail to detect objects inside a specified image region while keeping the rest of the image usable.