Research map / Image generation / GANs
Dual-Agent GANs for Photorealistic and Identity Preserving Profile Face Synthesis
Read the paperCheck 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 |
|---|---|---|---|
| Adding DA-GAN-refined synthetic profile faces to training data (vs. original data only) | deep recognition model fine-tuned on IJB-A · IJB-A verification | TAR @ FAR=0.001: +1.00% (0.920 to 0.930) | Helped |
| Adding raw simulator synthetic faces without DA-GAN refinement | deep recognition model fine-tuned on IJB-A · IJB-A verification | TAR @ FAR=0.001: -1.90% (0.920 to 0.901) | Hurt |
| Adding DA-GAN-refined faces vs. raw synthetic faces | deep recognition model on IJB-A · IJB-A identification | FNIR @ FPIR=0.01: 0.179 to 0.110 (vs b2); 0.125 to 0.110 (vs b1) | Helped |
| Adding DA-GAN-refined faces vs. original data only | deep recognition model on IJB-A · IJB-A identification | Rank-1: +0.50% (0.966 to 0.971) | Helped |
| DA-GAN refined data vs. NAN prior state of the art | DA-GAN recognition pipeline · IJB-A verification | TAR @ FAR=0.001: +4.90% (0.881 to 0.930) | Helped |
| Using DA-GAN (vs. ℓ2-softmax) on IJB-A verification at FAR=0.01 and 0.10 | DA-GAN recognition pipeline · IJB-A verification | TAR @ FAR=0.01: 0.976 vs 0.970 (DA-GAN higher); at FAR=0.001 0.930 vs 0.943 (DA-GAN lower) | Mixed |
| Ablations removing L_adv, L_ip, or L_pp from DA-GAN (ablation) | DA-GAN generator/discriminator · qualitative comparison | visual quality | No clear effect |
Closest papers by meaning
- FaceID-GAN: Learning a Symmetry Three-Player GAN for Identity-Preserving Face Synthesis (CVPR 2018 · similarity 0.92)
- Pose-Guided Photorealistic Face Rotation (CVPR 2018 · similarity 0.90)
- A 3D GAN for Improved Large-Pose Facial Recognition (CVPR 2021 · similarity 0.89)
- Disentangled Representation Learning GAN for Pose-Invariant Face Recognition (CVPR 2017 · similarity 0.89)
- Diffusion-driven GAN Inversion for Multi-Modal Face Image Generation (CVPR 2024 · similarity 0.87)
- Vec2Face: Scaling Face Dataset Generation with Loosely Constrained Vectors (ICLR 2025 · 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.