AI Paper Atlas

Research map / Detection and segmentation / CNN features

Language-Guided Hierarchical Fine-Grained Image Forgery Detection and Localization

Xiao Bo Guo, Xiaohong Liu, Iacopo Masi, Xiaoming Liu

IJCV 2024 · Detection and segmentation › CNN features · approach: Forgery detection · 38 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 LFLE (CLIP-based enhancer) to HiFi-Net (ablation)HiFi-Net · HiFi-IFDL, localizationAUC: 95.3 -> 95.7Improved
add manipulation score-map loss for LFLE (ablation)HiFi-Net++ · HiFi-IFDL, localizationAUC: 95.7 -> 97.7Improved
add LFLE to HiFi-Net, detection (ablation)HiFi-Net · HiFi-IFDL, detectionAUC: 96.8 -> 97.0No clear effect
use hierarchical path prediction instead of independent classification (ablation)HiFi-Net · HiFi-IFDL, detectionAUC: 93.2 -> 96.8Improved
classify at all four levels instead of only the 4th level (ablation)HiFi-Net · HiFi-IFDL, detectionAUC: 93.1 -> 96.8Improved
add classification module to localization-only model (ablation)HiFi-Net · HiFi-IFDL, detectionAUC: 65.0 -> 96.8Improved
add localization module to classification-only model (ablation)HiFi-Net · HiFi-IFDL, localizationAUC: 66.0 -> 95.3Improved
add LFLE (HiFi-Net -> HiFi-Net++)HiFi-Net · HiFi-IFDL image-editing domain, localizationAUC: 93.0 -> 97.8Improved
add LFLE (HiFi-Net -> HiFi-Net++)HiFi-Net · HiFi-IFDL overall detectionF1: AUC 96.8 -> 97.2, F1 94.1 -> 93.7Mixed
add LFLE (HiFi-Net -> HiFi-Net++)HiFi-Net · Pre-trained localization, 5 image-editing setsaverage AUC: 89.4 -> 91.2Improved
add LFLE (HiFi-Net -> HiFi-Net++)HiFi-Net · NIST16 robustness to post-processingaverage AUC: 84.2 -> 85.1Improved
add LFLE (HiFi-Net -> HiFi-Net++)HiFi-Net · Zero-shot GenAI inpainting localizationaverage AUC: .712 -> .730Improved

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