Research map / 3D vision / Face and shape
Face Forgery Detection by 3D Decomposition
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 direct light to identity texture (ablation) | VGG16 decomposition input · Faceforensics++ (FFpp) | AUC: 99.27 -> 99.56 | Helped |
| Normalize 3D shape by warping to UV space (ablation) | Identity texture + direct light input · Faceforensics++ (FFpp) | AUC: 99.29 -> 99.56 | Helped |
| Use ambient light and common texture without identity texture or direct light (ablation) | VGG16 decomposition input · Faceforensics++ (FFpp) | AUC: 99.13 -> 50.00 | Hurt |
| Fuse image and facial-detail streams instead of using one image stream (ablation) | FD²Net · DFD | AUC: 65.57 -> 78.65 | Helped |
| Replace halfway fusion with score fusion (ablation) | FD²Net · DFD | AUC: 78.65 -> 72.82 | Hurt |
| Replace feature fusion with halfway fusion (ablation) | FD²Net · DFD and DFDC | AUC: 78.55 -> 78.65; 65.36 -> 66.09 | Mixed |
| Add detail-difference-supervised attention to both streams (ablation) | FD²Net with halfway fusion · DFDC | AUC: 66.09 -> 67.70 | Helped |
| Remove facial-detail-difference supervision from attention (ablation) | FD²Net attention · DFDC | AUC: 67.70 -> 65.46 | Hurt |
| Add shape normalization to the facial-detail stream (ablation) | FD²Net · DFD | AUC: 72.51 -> 76.09 | Helped |
| Add texture normalization to the facial-detail stream (ablation) | FD²Net · DFD | AUC: 72.51 -> 76.84 | Helped |
| Combine shape and texture normalization (ablation) | FD²Net · DFD | AUC: 76.84 -> 79.08 | Helped |
| Use FD²Net rather than EfficientNetB4 Ensemble | Face forgery detection trained on FFpp · DFD (HQ) and DFDC | AUC: 72.82 -> 79.08; 63.03 -> 67.70 | Helped |
Closest papers by meaning
- Face X-Ray for More General Face Forgery Detection (CVPR 2020 · similarity 0.89)
- FakeDiffer: Distributional Disparity Learning on Differentiated Reconstruction for Face Forgery Detection (AAAI 2025 · similarity 0.86)
- Identity-Aware Vision-Language Model for Explainable Face Forgery Detection (AAAI 2026 · similarity 0.86)
- Face Forgery Video Detection via Temporal Forgery Cue Unraveling (CVPR 2025 · similarity 0.85)
- DevFD : Developmental Face Forgery Detection by Learning Shared and Orthogonal LoRA Subspaces (NeurIPS 2025 · similarity 0.84)
- Forensics Adapter: Adapting CLIP for Generalizable Face Forgery Detection (CVPR 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.