Research map / Image generation / GANs
Image Processing Using Multi-Code GAN Prior
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 |
|---|---|---|---|
| Use multiple latent codes with feature composition and adaptive channel importance instead of one latent code | PGGAN inversion · Bedroom, Church, and Face | PSNR and LPIPS: Bedroom PSNR: 18.34 -> 25.13; LPIPS: 0.5201 -> 0.1578; Church PSNR: 17.81 -> 22.76; LPIPS: 0.4789 -> 0.1799; Face PSNR: | Helped |
| Use multiple latent codes with feature composition and adaptive channel importance | PGGAN-based colorization · Bedroom and Church | AuC (%): Bedroom: 88.55 -> 90.02; Church: 89.13 -> 89.43 | Helped |
| Use mGANprior instead of specialized learned super-resolution methods | Face super-resolution, SR factor 16 · Face | PSNR, LPIPS, and NIQE: PSNR: 28.82 (RCAN) vs 26.93; LPIPS: 0.3862 (ESRGAN) -> 0.3584; NIQE: 3.27 (ESRGAN) -> 3.19 | Mixed |
| Use multiple latent codes with feature composition and adaptive channel importance | PGGAN-based inpainting · Center Crop and Random Crop | PSNR and SSIM: Center Crop PSNR: 17.92 -> 21.43; SSIM: 0.4327 -> 0.5320; Random Crop PSNR: 18.72 -> 22.11; SSIM: 0.2793 -> 0.5532 | Helped |
| Increase the number of latent codes (ablation) | mGANprior inversion · Image reconstruction | Reconstruction quality | Helped |
| Increase the number of latent codes beyond 20 (ablation) | mGANprior inversion · Image reconstruction | Reconstruction quality | No clear effect |
| Compose features at a higher generator layer (ablation) | mGANprior inversion · Image reconstruction | Reconstruction quality | Helped |
| Choose composition layers to match the task's representation needs (ablation) | mGANprior · Colorization and inpainting | Task performance | Mixed |
| Use multiple codes rather than a single code for a bedroom image from a mismatched GAN training domain (ablation) | GAN inversion · Bedroom image inverted using face, church, conference room, or bedroom GANs | Reconstruction quality | Helped |
Closest papers by meaning
- Posterior Promoted GAN With Distribution Discriminator for Unsupervised Image Synthesis (CVPR 2021 · similarity 0.89)
- High Fidelity GAN Inversion via Prior Multi-Subspace Feature Composition (AAAI 2021 · similarity 0.89)
- E$^2$GAN: Efficient Training of Efficient GANs for Image-to-Image Translation (ICML 2024 · similarity 0.87)
- Compressing Image-to-Image Translation GANs Using Local Density Structures on Their Learned Manifold (AAAI 2024 · similarity 0.87)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.87)
- RG-GAN: Dynamic Regenerative Pruning for Data-Efficient Generative Adversarial Networks (AAAI 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.