AI Paper Atlas

Research map / Video / Video segmentation and tracking

DREAM: A Benchmark Study for Deepfake PhotoRealism AssessMent

Bo Peng, Zichuan Wang, Yu Sheng, Xiaochuan Jin, Wei Wang, Jing Dong

TPAMI 2026 · Video › Video segmentation and tracking · approach: Deepfake video detection · 0 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 description-aligned CLIP training with textual branch to the OPDAI visual backbone and lossOPDAI backbone · DREAMavr: 0.782 -> 0.827Improved
pretrain on deepfake detection data instead of ImageNet (ablation)OPDAI · DREAMavr: 0.750 -> 0.782Improved
pretrain on deepfake detection data instead of ImageNet (ablation)UNILJ · DREAMavr: 0.611 -> 0.702Improved
replace MAE loss with RMSE loss (ablation)OPDAI · DREAMavr: 0.631 -> 0.725Improved
replace RMSE loss with KL loss (ablation)OPDAI · DREAMavr: 0.725 -> 0.810Improved
combine NinN with KL instead of KL alone (ablation)OPDAI · DREAMavr: 0.810 -> 0.782Hurt
add PLCC loss to RMSE (ablation)OPDAI · DREAMavr: 0.725 -> 0.788Improved
add rank loss to RMSE (ablation)OPDAI · DREAMavr: 0.725 -> 0.726No clear effect
use InternVL2.5-8B backbone instead of mPLUG-Owl2 (ablation)Q-Align · DREAMavr: 0.203 -> 0.644Improved
use InternVL2.5-8B backbone instead of mPLUG-Owl2 (ablation)DeQA-Score · DREAMavr: 0.213 -> 0.692Improved
use deepfake-detection features (DFGC1st) instead of ResNet50 featuresfeature-based regression · DREAMavr: 0.446 -> 0.682Improved
use top-11 retrieved descriptions instead of random-1 for explanation (ablation)DA-CLIP explanation · DREAM explanation errorRMSE (lower better): 0.099 -> 0.063Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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