Research map / Adversarial robustness and security / Attacks on text and graphs
3D-Aware Adversarial Makeup Generation for Facial Privacy Protection
Yueming Lyu, Yue Jiang, Ziwen He, Bo Peng, Yunfan Liu, Jing Dong
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 Makeup Adjustment Module (MAM) (ablation) | 3DAM-GAN · Makeup generation quality | FID: 31.52 -> 26.78 | Improved |
| add Makeup Adjustment Module (MAM) (ablation) | 3DAM-GAN · Makeup generation quality | PSNR: 19.63 -> 20.82 | Improved |
| add Makeup Adjustment Module (MAM) (ablation) | 3DAM-GAN · Makeup generation quality | SSIM: 0.90 -> 0.91 | Improved |
| add UV makeup loss (ablation) | 3DAM-GAN · Makeup generation quality | FID: 35.63 -> 26.78 | Improved |
| add UV makeup loss (ablation) | 3DAM-GAN · Makeup generation quality | PSNR: 19.29 -> 20.82 | Improved |
| add UV makeup loss (ablation) | 3DAM-GAN · Makeup generation quality | SSIM: 0.89 -> 0.91 | Improved |
| UV-space 3D-aware makeup generation instead of 2D makeup transfer (vs AMT-GAN) | 3DAM-GAN · CelebA-HQ black-box, FaceNet target | ASR at 0.01 FAR (%): 47.10 -> 64.33 | Improved |
| UV-space makeup generation vs AMT-GAN | 3DAM-GAN · CelebA-HQ black-box, IR152 target | ASR at 0.01 FAR (%): 36.27 -> 57.04 | Improved |
| UV-space makeup generation vs AMT-GAN | 3DAM-GAN · LADN black-box, MobileFace target | ASR at 0.01 FAR (%): 88.98 -> 88.65 | Hurt |
| UV-space makeup generation vs AMT-GAN | 3DAM-GAN · LFW face verification, FaceNet | PSR at 0.01 FAR (%): 26.68 -> 82.78 | Improved |
| UV-space makeup generation vs AMT-GAN | 3DAM-GAN · CelebA-HQ face identification, IRSE50 | Rank-1 protection success rate (%): 71.81 -> 95.52 | Improved |
| full-face UV makeup instead of eye-shadow patch (vs Adv-Makeup) in image quality | 3DAM-GAN · Makeup generation quality | FID: 3.05 -> 26.78 | Hurt |
Closest papers by meaning
- Protecting Facial Privacy: Generating Adversarial Identity Masks via Style-Robust Makeup Transfer (CVPR 2022 · similarity 0.93)
- Recoverable Facial Identity Protection via Adaptive Makeup Transfer Adversarial Attacks (AAAI 2025 · similarity 0.89)
- Anti-Makeup: Learning A Bi-Level Adversarial Network for Makeup-Invariant Face Verification (AAAI 2018 · similarity 0.89)
- DiffAM: Diffusion-based Adversarial Makeup Transfer for Facial Privacy Protection (CVPR 2024 · similarity 0.89)
- Natural Adversarial Mask for Face Identity Protection in Physical World (TPAMI 2024 · similarity 0.88)
- Non-Adaptive Adversarial Face Generation (NeurIPS 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.
