AI Paper Atlas

Research map / Detection and segmentation / CNN features

Exploring Disentangled Content Information for Face Forgery Detection

Jiahao Liang, Huafeng Shi, Weihong Deng

ECCV 2022 · Detection and segmentation › CNN features · approach: Forgery detection · 62 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 basic disentanglement framework alone (ablation)Xception · FF++-DFAUC: 95.36 -> 94.55Hurt
add Content Consistency Constraint (C2C) to basic disentanglement (ablation)Xception · Celeb-DFAUC: 65.11 -> 70.08Improved
add Global Representation Contrastive Constraint (GRCC) to basic disentanglement (ablation)Xception · Celeb-DFAUC: 65.11 -> 72.57Improved
combine C2C and GRCC (ablation)Xception · Celeb-DFAUC: 65.11 -> 76.91Improved
add the framework to ResNest-50ResNest-50 · Celeb-DF (trained on FF++-DF)AUC: 68.00 -> 82.38Improved
add the framework to XceptionXception · FF++ LQ in-dataset (AVG of 4 subsets)ACC: 89.79 -> 90.98Improved
add the framework to Xception, cross-method generalizationXception · FF++ C40 cross-method (train NT, mean AUC)AUC: 70.53 -> 70.01Mixed
feature-level random erasing augmentation instead of image-level (ablation)disentangled framework · FF++-DFDelta AUC: -2.13 -> +0.34Improved
feature-level horizontal flip instead of image-level (ablation)disentangled framework · Celeb-DFDelta AUC: +0.23 -> +3.62Improved
feature-level Mixup instead of image-level (ablation)disentangled framework · FF++-DFDelta AUC: -0.85 -> -0.07Improved
add the framework to Xception on identity-unbalanced FF++Xception · ID unbalanced FF++AUC: 96.91 -> 99.32Improved
add the framework to Xception on background-unbalanced FF++Xception · BG unbalanced FF++AUC: 92.42 -> 98.71Improved

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