AI Paper Atlas

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

Implicit Concept Removal of Diffusion Models

Zhili Liu, Kai Chen, Yifan Zhang, Jianhua Han, Lanqing Hong, Hang Xu, Zhenguo Li, Dit-Yan Yeung, James Kwok

ECCV 2024 · Image generation › Text-to-image diffusion · approach: Concept erasure in diffusion models · 27 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
add geometric location tokens and concept name to captions, used as negative promptStable Diffusion v1-5 · I2P overallinappropriate ratio: 0.39 -> 0.09Improved
apply Geom-Erasing in data removal fine-tuningSD fine-tuned on ICD-QR · ICD-QRICR (%): 74.59 -> 5.38Improved
apply Geom-Erasing in data removal fine-tuningSD fine-tuned on ICD-Text · ICD-TextICR (%): 71.84 -> 13.48Improved
erasing implicit concepts also improves FIDSD fine-tuned on ICD-QR · ICD-QRFID: 65.82 -> 41.41Improved
use ESD for implicit concept removalSD fine-tuned on ICD-QR · ICD-QRFID: 65.82 -> 90.97Hurt
use FMN or NP or SLD for implicit concept removalSD · ICD-QRICR (%): 74.59 -> 80.42 (FMN); 59.64 (NP)No clear effect
add geometry bins (selected bin number 72 at 32^2) versus smaller or larger bin settings (ablation)Geom-Erasing · ICD-WatermarkICR (%): ICR 7.36 default vs 18.26 (8^2) and 16.47 (64^2)Mixed
use detector-confidence re-weighting instead of fixed weight (ablation)Geom-Erasing · ICD-WatermarkICR (%): 7.23 -> 9.21Hurt
add concept name and uniform geometry to negative prompt (ablation)Geom-Erasing · ICD-WatermarkICR (%): 7.23 -> 5.02Improved
add geometric information to negative prompt (ablation)Geom-Erasing · ICD-WatermarkFID: 6.15 -> 6.99Hurt
add noise to location tokens (geometry accuracy below 0.4 IoU) (ablation)Geom-Erasing · ICD-WatermarkICRHurt
include the word watermark in the text promptStable Diffusion · watermark presence over 1000 samplescorrelation with watermark presence: r=-0.08, p=0.21No clear effect

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