Research map / 3D vision / Face and shape
FSGAN: Subject Agnostic Face Swapping and Reenactment
Yuval Nirkin, Yosi Keller, Tal Hassner
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 completion network Gc to Gr (ablation) | FSGAN (Gr) · FaceForensics++ ablation (first 100 frames of 500 videos) | pose Euler error: 3.16 -> 3.21 | Hurt |
| add completion network Gc to Gr (ablation) | FSGAN (Gr) · FaceForensics++ ablation (first 100 frames of 500 videos) | landmark error: 22.6 -> 24.5 | Hurt |
| add blending network Gb to Gr (ablation) | FSGAN (Gr) · FaceForensics++ ablation (first 100 frames of 500 videos) | pose Euler error: 3.16 -> 2.75 | Improved |
| add blending network Gb to Gr (ablation) | FSGAN (Gr) · FaceForensics++ ablation (first 100 frames of 500 videos) | SSIM: 0.54 -> 0.52 | Hurt |
| add completion network Gc to Gr+Gb (ablation) | FSGAN (Gr+Gb) · FaceForensics++ ablation (first 100 frames of 500 videos) | pose Euler error: 2.75 -> 2.49 | Improved |
| add completion network Gc to Gr+Gb (ablation) | FSGAN (Gr+Gb) · FaceForensics++ ablation (first 100 frames of 500 videos) | landmark error: 23.6 -> 22.2 | Improved |
| add blending network Gb to Gr+Gc (ablation) | FSGAN (Gr+Gc) · FaceForensics++ ablation (first 100 frames of 500 videos) | pose Euler error: 3.21 -> 2.49 | Improved |
| add blending network Gb to Gr+Gc (ablation) | FSGAN (Gr+Gc) · FaceForensics++ ablation (first 100 frames of 500 videos) | SSIM: 0.54 -> 0.51 | Hurt |
| use FSGAN instead of Nirkin et al. | Nirkin et al. · FaceForensics++ swapping (first 100 frames of 500 videos) | pose Euler error: 3.15 -> 2.49 | Improved |
| use FSGAN instead of Nirkin et al. | Nirkin et al. · FaceForensics++ swapping (first 100 frames of 500 videos) | landmark error: 26.5 -> 22.2 | Improved |
| use FSGAN instead of DeepFakes | DeepFakes · FaceForensics++ swapping (first 100 frames of 500 videos) | pose Euler error: 4.05 -> 2.49 | Improved |
| use FSGAN instead of DeepFakes | DeepFakes · FaceForensics++ swapping (first 100 frames of 500 videos) | SSIM: 0.50 -> 0.51 | No clear effect |
Later papers that compared with this method
2 later accepted papers report FSGAN as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Human-Inspired Facial Sketch Synthesis with Dynamic Adaptation (ICCV 2023 · better on 2 of 3 shared results)
- FSGANv2: Improved Subject Agnostic Face Swapping and Reenactment (TPAMI 2022 · better on 2 of 3 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- FSGANv2: Improved Subject Agnostic Face Swapping and Reenactment (TPAMI 2022 · similarity 0.98)
- One-Shot Neural Face Reenactment via Finding Directions in GAN’s Latent Space (IJCV 2024 · similarity 0.89)
- Fine-Grained Face Swapping via Regional GAN Inversion (CVPR 2023 · similarity 0.89)
- Fast Face-Swap Using Convolutional Neural Networks (ICCV 2017 · similarity 0.88)
- ReenactGAN: Learning to Reenact Faces via Boundary Transfer (ECCV 2018 · similarity 0.88)
- FSRT: Facial Scene Representation Transformer for Face Reenactment from Factorized Appearance Head-pose and Facial Expression Features (CVPR 2024 · similarity 0.86)
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.
