Research map / Image generation / Editing and style transfer
Frequency-domain Manipulation for Face Obfuscation
Jintae Kim, Keunsoo Ko, Chang-Su Kim
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 perturbation module on LH and HL subbands (ablation) | FreM (neutralization only) · LFW face recognition | accuracy (%): 99.15 -> 99.61 | Improved |
| add suppression module replacing HH subband (ablation) | FreM (neutralization + perturbation) · LFW reconstruction attack | reconstruction PSNR (dB): 16.72 -> 13.23 | Improved |
| add suppression module replacing HH subband (ablation) | FreM (neutralization + perturbation) · LFW face recognition | accuracy (%): 99.61 -> 99.50 | Hurt |
| use average face C_LL for neutralization (ablation) | FreM · LFW face recognition | accuracy (%): 98.13 -> 99.50 | Improved |
| set block size P to 8 instead of 112 (ablation) | FreM · LFW face recognition | accuracy (%): 99.27 -> 99.52 | Improved |
| set LL partition size to 4 instead of 1 (ablation) | FreM · LFW face recognition | accuracy (%): 98.68 -> 99.52 | Improved |
| add coefficient energy constraint loss with weight 0.01 (ablation) | FreM · LFW face recognition | accuracy (%): 99.37 -> 99.53 | Improved |
| add coefficient energy constraint loss (ablation) | FreM · LFW reconstruction attack | qualitative reconstruction | Improved |
| frequency-domain DCT manipulation instead of spatial filters (versus Forbes) | Forbes · LFW ODR | accuracy (%): 82.77 -> 98.67 | Improved |
| multi-task joint optimization versus age-only on age feature distance (ablation) | FreM · Faceptor multi-task feature distance | age feature distance (lower is better): 1.81 -> 1.82 | Mixed |
Closest papers by meaning
- Privacy-Preserving Face Recognition in the Frequency Domain (AAAI 2022 · similarity 0.88)
- Transferable Facial Privacy Protection against Blind Face Restoration via Domain-Consistent Adversarial Obfuscation (ICML 2024 · similarity 0.87)
- Disguise without Disruption: Utility-Preserving Face De-identification (AAAI 2024 · similarity 0.84)
- How Private is Private? A Comparative Study for Face De-Identification (NeurIPS 2026 · similarity 0.83)
- WaveFace: Authentic Face Restoration with Efficient Frequency Recovery (CVPR 2024 · similarity 0.83)
- CryptoFace: End-to-End Encrypted Face Recognition (CVPR 2025 · 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.
