Research map / Federated learning and privacy / Privacy attacks and secure learning
Learning a Dynamic Privacy-preserving Camera Robust to Inversion Attacks
Jiacheng Cheng, Xiang Dai, Jia Wan, Nick Antipa, Nuno Vasconcelos
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 a dynamic (randomly sampled) PSF so attackers cannot recover it, versus static PSF under white-box inversion | DyPP camera · PubFig face recognition after TVD attack | accuracy: 54% (white-box) -> 14% (black-box) | Improved |
| PSF inversion with true PSF instead of blind or deep-learning deblurring | static-PSF cameras · Reconstruction PSNR/SSIM | PSNR: from ~15 to ~20 PSNR, ~0.55 to ~0.75 SSIM | Improved |
| DyPP versus PP-HPE lens on object detection | DyPP · COCO Detection val | AP: 33.9 -> 36.0 | Improved |
| DyPP versus PP-HPE lens on pose estimation | DyPP · COCO Keypoints val | AP: 42.8 -> 42.2 | No clear effect |
| DyPP versus PP-HPE lens on crowd counting | DyPP · ShanghaiTech B | MAE: 21.3 -> 19.3 | Improved |
| use extremely low-resolution sensor for privacy | Low-resolution camera · COCO pose / detection | AP: pose AP 10.3 vs 42.2 for DyPP | Hurt |
| black-box attack with 64 sampled PSFs instead of white-box | DyPP · Reconstruction PSNR | PSNR: 20.3 -> 17.3 | Improved |
Closest papers by meaning
- Learning Privacy-Preserving Optics for Human Pose Estimation (ICCV 2021 · similarity 0.88)
- Validating Privacy-Preserving Face Recognition under a Minimum Assumption (CVPR 2024 · similarity 0.87)
- Asymmetric Invertible Threat: Learning Reversible Privacy Defense for Face Recognition (NeurIPS 2026 · similarity 0.85)
- Privacy-Preserving Face Recognition Using Trainable Feature Subtraction (CVPR 2024 · similarity 0.84)
- Privacy-Preserving Optics for Enhancing Protection in Face De-Identification (CVPR 2024 · similarity 0.84)
- Transferable Facial Privacy Protection against Blind Face Restoration via Domain-Consistent Adversarial Obfuscation (ICML 2024 · similarity 0.83)
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.
