AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

LoRC: Detecting AI-Generated Images via Low-Rank Collapse in the Semantic-Residual Space

Haozhen Yan, Ruoxin Chen, Jiahui Zhan, Taiping Yao, Bo Wang, Youchang xiao, Shouhong Ding, Liqing Zhang, Jianfu Zhang

ECCV 2026 · Image generation › Autoregressive and latent generation · approach: Generalizable fake image detection

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 orthogonal semantic decomposition to baseline (ablation)LoRC · T2I-CoReBenchbalanced accuracy (%): 94.0 -> 98.1Improved
add orthogonal semantic decomposition to baseline (ablation)LoRC · In-the-wildbalanced accuracy (%): 89.4 -> 86.9Hurt
add low-rank attention after decomposition (ablation)LoRC · In-the-wildbalanced accuracy (%): 86.9 -> 90.2Improved
add low-rank attention after decomposition (ablation)LoRC · Standard benchmarksbalanced accuracy (%): 97.2 -> 96.9No clear effect
add Subspace Separation Loss (ablation)LoRC · Average (Standard, In-the-wild, T2I)balanced accuracy (%): 94.4 -> 96.8Improved
apply low-rank attention to original entangled patch features (ablation)LoRC variant · Averagebalanced accuracy (%): 93.3 -> 90.0Hurt
use orthogonal projection instead of subtraction for decomposition (ablation)LoRC · Averagebalanced accuracy (%): 96.0 -> 96.8Improved
add frozen [CLS] token to residual features (ablation)LoRC · Averagebalanced accuracy (%): 96.2 -> 96.8Improved
place SSL after low-rank bottleneck instead of before (ablation)LoRC · Averagebalanced accuracy (%): 95.0 -> 96.8Improved
scale backbone from DINOv3-L to DINOv3-7B (ablation)LoRC · Averagebalanced accuracy (%): 93.6 -> 97.4Improved
use LoRC (decomposition + low-rank + SSL) instead of original patch features (ablation)DINOv3-H+ · Averagebalanced accuracy (%): 93.3 -> 96.8Improved
LoRC vs DDA on unseen T2I generatorsLoRC · T2I-CoReBenchbalanced accuracy (%): 91.1 -> 97.0Improved

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