AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Seeing is not always believing: Benchmarking Human and Model Perception of AI-Generated Images

Zeyu Lu, Di Huang, LEI BAI, Jingjing Qu, Chengyue Wu, Xihui Liu, Wanli Ouyang

NeurIPS 2023 · Image generation › Autoregressive and latent generation · approach: Generalizable fake image detection · 31 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
train on diverse mixed dataset (Setting D) instead of single-generator Setting B (ablation)ConvNext-S (B+J 0.5) · MPBench total average accuracyaccuracy (%): 55.2 -> 83.0Improved
train on diverse mixed dataset (Setting D) instead of Setting C (ablation)ResNet50 (B+J 0.5) · MPBench total average accuracyaccuracy (%): 47.5 -> 79.4Improved
train on diverse mixed dataset (Setting D) instead of Setting A (ablation)CLIP-ViT-L (LC) · MPBench total average accuracyaccuracy (%): 82.0 -> 81.6No clear effect
use CLIP-ViT-L frozen backbone vs ConvNext fine-tuned (ablation)detector backbone · MPBench Setting Atotal average accuracy (%): 70.2 -> 82.0Improved
use ConvNext vs CLIP-ViT-L for real image accuracy (ablation)detector backbone · MPBench Setting D real averageaccuracy (%): 75.4 -> 99.1Improved
blur+JPEG augmentation probability 0.5 vs 0.1 (ablation)ConvNext-S · MPBench Setting D totalaccuracy (%): 81.6 -> 83.0Improved
blur+JPEG augmentation probability 0.5 vs 0.1 (ablation)ConvNext-S · MPBench Setting A totalaccuracy (%): 70.2 -> 68.6Hurt
participants with AIGC background vs withouthumans · HPBenchaccuracy: +2.7% overall; +0.7% real; +3.7% fakeMixed
object category vs multiperson categoryhumans · HPBenchaccuracy: 0.6750 -> 0.5083Hurt

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