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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| train on diverse mixed dataset (Setting D) instead of single-generator Setting B (ablation) | ConvNext-S (B+J 0.5) · MPBench total average accuracy | accuracy (%): 55.2 -> 83.0 | Improved |
| train on diverse mixed dataset (Setting D) instead of Setting C (ablation) | ResNet50 (B+J 0.5) · MPBench total average accuracy | accuracy (%): 47.5 -> 79.4 | Improved |
| train on diverse mixed dataset (Setting D) instead of Setting A (ablation) | CLIP-ViT-L (LC) · MPBench total average accuracy | accuracy (%): 82.0 -> 81.6 | No clear effect |
| use CLIP-ViT-L frozen backbone vs ConvNext fine-tuned (ablation) | detector backbone · MPBench Setting A | total average accuracy (%): 70.2 -> 82.0 | Improved |
| use ConvNext vs CLIP-ViT-L for real image accuracy (ablation) | detector backbone · MPBench Setting D real average | accuracy (%): 75.4 -> 99.1 | Improved |
| blur+JPEG augmentation probability 0.5 vs 0.1 (ablation) | ConvNext-S · MPBench Setting D total | accuracy (%): 81.6 -> 83.0 | Improved |
| blur+JPEG augmentation probability 0.5 vs 0.1 (ablation) | ConvNext-S · MPBench Setting A total | accuracy (%): 70.2 -> 68.6 | Hurt |
| participants with AIGC background vs without | humans · HPBench | accuracy: +2.7% overall; +0.7% real; +3.7% fake | Mixed |
| object category vs multiperson category | humans · HPBench | accuracy: 0.6750 -> 0.5083 | Hurt |
Closest papers by meaning
- Is Artificial Intelligence Generated Image Detection a Solved Problem? (NeurIPS 2025 · similarity 0.87)
- Semi-Truths: A Large-Scale Dataset of AI-Augmented Images for Evaluating Robustness of AI-Generated Image detectors (NeurIPS 2024 · similarity 0.87)
- Human and AI Perceptual Differences in Image Classification Errors (AAAI 2025 · similarity 0.87)
- BIAS-ID: A Framework for Analyzing Transformation Biases in AI-Generated Image Detectors (NeurIPS 2026 · similarity 0.86)
- Community Forensics: Using Thousands of Generators to Train Fake Image Detectors (CVPR 2025 · similarity 0.86)
- Are High-Quality AI-Generated Images More Difficult for Models to Detect? (ICML 2025 · similarity 0.86)
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.
