AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

Towards Human-Imperceptible Backdoor Attacks on Text-to-Image Diffusion Models

Yiming Wu, Chenghao Chen, Changkun Wu, Chong Fu, Biru Zhu, Zhenyu Wen, Zhen Hong

CVPR 2026 · Image generation › Text-to-image diffusion · approach: Inference-time T2I improvement

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
add cover samples (keyword-only and syntax-only captions) to poisoned data (ablation)clean-label backdoor · woman->nude womanFTR-W (%): 80.1 -> 1.9Improved
add cover samples (keyword-only and syntax-only captions) to poisoned data (ablation)clean-label backdoor · woman->nude womanFTR-S (%): 86.6 -> 4.7Improved
add syntax-based cover samples to keyword-only covers (ablation)clean-label backdoor · woman->nude womanFTR-W (%): 55.0 -> 1.9Improved
add keyword cover samples to syntax-only covers (ablation)clean-label backdoor · woman->nude womanASR-N (%): 59.8 -> 98.0Improved
use composite trigger (synonym plus sentence structure) instead of a single-modification trigger (ablation)text trigger · prompts with low commonnessfalse trigger rateImproved
increase poisoned samples from 10 to 30 (ablation)clean-label backdoor · woman->nude womanASR-H (%): 0 -> 87.5Improved
increase poisoned samples from 50 to 70 (ablation)clean-label backdoor · woman->nude womanASR-H (%): 99.4 -> 99.9No clear effect
use a clean-label rather than dirty-label poisoningtext-to-image backdoor · 3 attack scenarios averageUCR (%): 100 -> 0Improved
use clean-label rather than dirty-label poisoning (Wu et al.)text-to-image backdoor · 3 attack scenarios averageFTR (%): 19.5 -> 5.1Improved
use clean-label rather than dirty-label poisoning (Huang et al.)text-to-image backdoor · 3 attack scenarios averageASR-N (%): 99.9 -> 90.4Hurt
apply Gaussian noise preprocessing defense at 0.03 or higherclean-label backdoor · woman->nude womanASR-H (%): 100 -> 0Hurt

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