Research map / Federated learning and privacy / Privacy attacks and secure learning
PrivateEyes: Gaze-Preserving Anonymization for Data Sharing
Surabhi Gupta, Dinesh Prabhu Muthumariappan, Biplab Das, Anoop Kolar Rajagopal, Kiran Nanjunda Iyer, Donghwan Seo
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 pose estimation network (PEN) pose embedding to ControlEye (ablation) | PrivateEyes · OpenEDS2019 | gaze error (deg): 4.94 -> 4.43 | Improved |
| add PEN pose embedding to ControlEye (ablation) | PrivateEyes · EV-Eye | gaze error (deg): 3.33 -> 2.89 | Improved |
| add PEN pose embedding to ControlEye (ablation) | PrivateEyes · LPW | gaze error (deg): 3.27 -> 2.51 | Improved |
| add PEN pose embedding to ControlEye (ablation) | PrivateEyes · OpenEDS2019 | iris recognition accuracy, cropped (%): 16.1 -> 16.5 | No clear effect |
| add PEN pose embedding to ControlEye (ablation) | PrivateEyes · OpenEDS2019 | FID: 68.11 -> 69.25 | No clear effect |
| use diffusion-based PrivateEyes instead of Iris Style Transfer | eye anonymization · OpenEDS2019 | gaze error (deg): 7.65 -> 4.43 | Improved |
| use PrivateEyes instead of Rubber Sheet Model | eye anonymization · OpenEDS2019 | iris recognition accuracy, full iris (%): 87.0 -> 14.2 | Improved |
| use PrivateEyes instead of Iris Style Transfer | eye anonymization · LPW | gaze error (deg): 1.91 -> 2.51 | Hurt |
| use PrivateEyes instead of simple blur/noise/downsample | eye anonymization · OpenEDS2019 | gaze error (deg): 3.88 (blur) vs 4.43 | Mixed |
| use PrivateEyes instead of Iris Style Transfer | eye anonymization · OpenEDS2019 | LPIPS: 0.181 -> 0.272 | Hurt |
Closest papers by meaning
- Privacy-Preserving Optics for Enhancing Protection in Face De-Identification (CVPR 2024 · similarity 0.84)
- MYOPIA: Protecting Face Privacy from Malicious Personalized Text-to-Image Synthesis via Unlearnable Examples (AAAI 2025 · similarity 0.84)
- EgoPrivacy: What Your First-Person Camera Says About You? (ICML 2025 · similarity 0.83)
- Validating Privacy-Preserving Face Recognition under a Minimum Assumption (CVPR 2024 · similarity 0.83)
- PerceptAnon: Exploring the Human Perception of Image Anonymization Beyond Pseudonymization for GDPR (ICML 2024 · similarity 0.83)
- ReGenHuman: Re-Generating Human Appearances for Realistic Full-Body Video Anonymization (NeurIPS 2026 · 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.
