Research map / Image generation / Editing and style transfer
Beyond Face Rotation: Global and Local Perception GAN for Photorealistic and Identity Preserving Frontal View Synthesis
Rui Huang, Shu Zhang, Tianyu Li, Ran He
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 local pathway (ablation) | TP-GAN · MultiPIE Setting 2 (+-90 degrees) | Rank-1 (%): 44.13 -> 64.64 | Improved |
| add the identity preserving loss (ablation) | TP-GAN · MultiPIE Setting 2 (+-90 degrees) | Rank-1 (%): 43.23 -> 64.64 | Improved |
| add the adversarial loss (ablation) | TP-GAN · MultiPIE Setting 2 (+-90 degrees) | Rank-1 (%): 62.83 -> 64.64 | Improved |
| add the symmetry loss (ablation) | TP-GAN · MultiPIE Setting 2 (+-90 degrees) | Rank-1 (%): 62.47 -> 64.64 | Improved |
| add the local pathway (ablation) | TP-GAN · MultiPIE Setting 2 (+-60 degrees) | Rank-1 (%): 80.64 -> 87.72 | Improved |
| add the identity preserving loss (ablation) | TP-GAN · MultiPIE Setting 2 (+-60 degrees) | Rank-1 (%): 70.99 -> 87.72 | Improved |
| add the adversarial loss (ablation) | TP-GAN · MultiPIE Setting 2 (+-15 degrees) | Rank-1 (%): 98.09 -> 98.68 | Improved |
| add the symmetry loss (ablation) | TP-GAN · MultiPIE Setting 2 (+-15 degrees) | Rank-1 (%): 98.47 -> 98.68 | Improved |
| frontalize with TP-GAN before recognition versus raw profile images with Light CNN | Light CNN features · MultiPIE Setting 1 (+-90 degrees) | Rank-1 (%): 9.00 -> 64.03 | Improved |
| frontalize with TP-GAN before recognition | Light CNN features · MultiPIE Setting 1 (+-15 degrees) | Rank-1 (%): 99.78 -> 99.78 | No clear effect |
| frontalize with TP-GAN before recognition | Light CNN features · MultiPIE Setting 2 (+-90 degrees) | Rank-1 (%): 5.51 -> 64.64 | Improved |
| use TP-GAN synthesized images instead of original profile images | gender classifier · MultiPIE Setting 1 gender (+-45 degrees) | accuracy (%): 86.22 -> 90.71 | Improved |
Closest papers by meaning
- Dual-Agent GANs for Photorealistic and Identity Preserving Profile Face Synthesis (NeurIPS 2017 · similarity 0.91)
- 3D-Aided Dual-Agent GANs for Unconstrained Face Recognition (TPAMI 2018 · similarity 0.89)
- Pose-Guided Photorealistic Face Rotation (CVPR 2018 · similarity 0.89)
- FaceID-GAN: Learning a Symmetry Three-Player GAN for Identity-Preserving Face Synthesis (CVPR 2018 · similarity 0.89)
- Unsupervised Face Normalization With Extreme Pose and Expression in the Wild (CVPR 2019 · similarity 0.88)
- Diffusion-driven GAN Inversion for Multi-Modal Face Image Generation (CVPR 2024 · similarity 0.87)
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.
