AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

When Visual Privacy Protection Meets Multimodal Large Language Models

Xiaofei Hui, Qian Wu, Haoxuan Qu, Majid Mirmehdi, Hossein Rahmani, Jun Liu

IJCV 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: Vision-language model robustness · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use augmented Tchebycheff objective instead of weighted sum (ablation)Video-LLaVA anonymizer · UCF101-VISPRAction Top-1: 32.9 -> 48.4Improved
use augmented Tchebycheff objective instead of weighted sum (ablation)Video-LLaVA anonymizer · UCF101-VISPRPrivacy cMAP (lower better): 58.4 -> 54.6Improved
add augmentation term to the Tchebycheff norm (ablation)Video-LLaVA anonymizer · UCF101-VISPRAction Top-1: 42.6 -> 48.4Improved
add augmentation term to the Tchebycheff norm (ablation)Video-LLaVA anonymizer · UCF101-VISPRPrivacy cMAP (lower better): 57.3 -> 54.6Improved
add history gradients to SPSA estimation (ablation)Video-LLaVA anonymizer · UCF101-VISPRAction Top-1: 35.7 -> 48.4Improved
add history gradients to SPSA estimation (ablation)Video-LLaVA anonymizer · UCF101-VISPRPrivacy cMAP (lower better): 57.5 -> 54.6Improved
add critical gradient collection by L2 norm (ablation)Video-LLaVA anonymizer · UCF101-VISPRAction Top-1: 44.2 -> 48.4Improved
add gradient direction improvement with history (ablation)Video-LLaVA anonymizer · UCF101-VISPRAction Top-1: 43.4 -> 48.4Improved
add adaptive step size adjustment (ablation)Video-LLaVA anonymizer · UCF101-VISPRAction Top-1: 43.9 -> 48.4Improved
use critical-history optimization instead of SPSA (ablation)Video-LLaVA anonymizer · UCF101-VISPRMLLM calls per sample: 136 -> 24Improved
use critical-history optimization instead of SPSA-GC (ablation)Video-LLaVA anonymizer · UCF101-VISPRAction Top-1: 37.9 -> 48.4Improved
use critical-history optimization instead of evolution strategies (ablation)Video-LLaVA anonymizer · UCF101-VISPRAction Top-1: 30.2 -> 48.4Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy