Research map / Video / Video generation
Live Face De-Identification in Video
Oran Gafni, Lior Wolf, Yaniv Taigman
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 the face descriptor concatenated to the latent code (ablation) | Ours (low-res) · Face-descriptor ablation (qualitative) | visual quality / artifacts | Improved |
| add the mask output (ablation) | Ours (low-res) · Ablation, pixel vs ID distance | face edge quality / occlusion handling | Improved |
| apply adversarial loss to both raw and masked outputs (ablation) | Ours (low-res) · Ablation, pixel vs ID distance | visual artifacts | Improved |
| add gradual increase of lambda during training (ablation) | Ours (low-res) · Ablation, pixel vs ID distance | naturalness of generated face | Improved |
| add a lower-resolution output to the compound loss (ablation) | Ours (low-res) · Ablation, pixel vs ID distance | de-identification strength / artifacts | Hurt |
| add mask regularization (alpha 4 and 5) (ablation) | Ours · Mask regularization ablation (qualitative) | artifacts around occlusions and teeth | Improved |
| switch from lower-res to higher-res model | Ours · ArcFace rank of true identity (54,000 persons), average | median rank of true identity: 3773 -> 2776 | Mixed |
| apply de-identification to LFW image pairs | FaceNet · LFW pairs, FaceNet | true positive rate: 0.99 -> less than 0.04 | Improved |
Closest papers by meaning
- Face Video Deblurring Using 3D Facial Priors (ICCV 2019 · similarity 0.90)
- Face De-Identification: A Domain-Centric Survey From Capture to Processing (TPAMI 2026 · similarity 0.87)
- FaceMe: Robust Blind Face Restoration with Personal Identification (AAAI 2025 · similarity 0.86)
- RED-Net: A Recurrent Encoder–Decoder Network for Video-Based Face Alignment (IJCV 2018 · similarity 0.86)
- Disguise without Disruption: Utility-Preserving Face De-identification (AAAI 2024 · similarity 0.84)
- Enhancing Identity-Deformation Disentanglement in StyleGAN for One-Shot Face Video Re-Enactment (AAAI 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.
