Research map / Detection and segmentation / Open-set and few-shot detection
All in One: Unifying Deepfake Detection, Tampering Localization, and Source Tracing with a Robust Landmark-Identity Watermark
Junjiang Wu, Liejun Wang, Zhiqing Guo
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 |
|---|---|---|---|
| add landmark regression loss (lambda_L) (ablation) | LIDMark (FHD) · CelebA-HQ 128x128 common distortions | AED: 342.39 -> 1.17 | Improved |
| add landmark regression loss (lambda_L) (ablation) | LIDMark (FHD) · CelebA-HQ 128x128 deepfake manipulations | AED: 359.98 -> 1.58 | Improved |
| add landmark regression loss (lambda_L) (ablation) | LIDMark (FHD) · CelebA-HQ 256x256 common distortions | AED: 540.91 -> 3.06 | Improved |
| add landmark regression loss (lambda_L) (ablation) | LIDMark (FHD) · CelebA-HQ 128x128 deepfake manipulations | BER: 0.14% -> 3.33% | Hurt |
| add identifier loss (lambda_ID) (ablation) | LIDMark (FHD) · CelebA-HQ 128x128 common distortions | BER: 0.42% -> 0.27% | Improved |
| add identifier loss (lambda_ID) (ablation) | LIDMark (FHD) · CelebA-HQ 128x128 deepfake manipulations | BER: 0.49% -> 3.33% | Hurt |
| add identifier loss (lambda_ID) (ablation) | LIDMark (FHD) · CelebA-HQ 128x128 common distortions | AED: 2.31 -> 1.17 | Improved |
| add identifier loss (lambda_ID) (ablation) | LIDMark (FHD) · CelebA-HQ 128x128 | PSNR: 47.31 -> 40.22 | Hurt |
| add landmark regression loss (lambda_L) (ablation) | LIDMark (FHD) · CelebA-HQ 128x128 | PSNR: 42.12 -> 40.22 | Hurt |
| expand identifier payload from 16-D to 32-D (168-D total) (ablation) | LIDMark (FHD) · CelebA-HQ 256x256 | PSNR: 44.31 -> 42.36 | Hurt |
| expand identifier payload from 16-D to 32-D (168-D total) (ablation) | LIDMark (FHD) · CelebA-HQ 256x256 deepfake manipulations | AED: 4.01 -> 3.01 | Improved |
| expand identifier payload from 16-D to 32-D (168-D total) (ablation) | LIDMark (FHD) · CelebA-HQ 256x256 deepfake manipulations | BER: 2.55% -> 3.26% | Hurt |
Closest papers by meaning
- ForensicHub: A Unified Benchmark & Codebase for All-Domain Fake Image Detection and Localization (NeurIPS 2025 · similarity 0.85)
- DF40: Toward Next-Generation Deepfake Detection (NeurIPS 2024 · similarity 0.85)
- Exposing the Deception: Uncovering More Forgery Clues for Deepfake Detection (AAAI 2024 · similarity 0.85)
- Forensics Adapter: Adapting CLIP for Generalizable Face Forgery Detection (CVPR 2025 · similarity 0.84)
- OmniGuard: Hybrid Manipulation Localization via Augmented Versatile Deep Image Watermarking (CVPR 2025 · similarity 0.84)
- Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detector (CVPR 2025 · similarity 0.84)
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.
