Research map / Image generation / GANs
Deepfake Network Architecture Attribution
Tianyun Yang, Ziyao Huang, Juan Cao, Lei Li, Xirong Li
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 |
|---|---|---|---|
| use DNA-Det rather than LeveFreq | deepfake architecture attribution · celebA cross-dataset | accuracy: 47.07 -> 94.95 | Improved |
| use DNA-Det rather than LeveFreq | deepfake architecture attribution · LSUN-bedroom cross-dataset | accuracy: 53.42 -> 83.45 | Improved |
| add patchwise contrastive learning (ablation) | DNA-Det base network · celebA cross-dataset | accuracy: 33.02 -> 89.93 | Improved |
| add image-transformation pre-training (ablation) | DNA-Det base network · celebA cross-dataset | accuracy: 33.02 -> 53.00 | Improved |
| add patchwise contrastive learning (ablation) | DNA-Det base network · LSUN-bedroom cross-dataset | accuracy: 21.66 -> 81.09 | Improved |
| add image-transformation pre-training (ablation) | DNA-Det base network · LSUN-bedroom cross-dataset | accuracy: 21.66 -> 49.73 | Improved |
| add patchwise contrastive learning (ablation) | DNA-Det base network · GANs in the wild cross-dataset | accuracy: 47.95 -> 92.60 | Improved |
| add image-transformation pre-training (ablation) | DNA-Det base network · GANs in the wild cross-dataset | accuracy: 47.95 -> 73.02 | Improved |
| increase resize to 512px and use 1/8×1/8 patch crop ratio (ablation) | patchwise contrastive learning · GANs in the wild cross-dataset | accuracy: 21.04 -> 94.95 | Improved |
Closest papers by meaning
- Artificial Fingerprinting for Generative Models: Rooting Deepfake Attribution in Training Data (ICCV 2021 · similarity 0.90)
- TraceEvader: Making DeepFakes More Untraceable via Evading the Forgery Model Attribution (AAAI 2024 · similarity 0.89)
- Attributing Fake Images to GANs: Learning and Analyzing GAN Fingerprints (ICCV 2019 · similarity 0.89)
- Towards Discovery and Attribution of Open-World GAN Generated Images (ICCV 2021 · similarity 0.88)
- Towards Universal Fake Image Detectors That Generalize Across Generative Models (CVPR 2023 · similarity 0.88)
- Transfer Learning of Real Image Features with Soft Contrastive Loss for Fake Image Detection (AAAI 2025 · similarity 0.85)
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.
