Image Super Resolution On Ffhq 512 X 512 4X
Metrics
FED
FID
LLE
LPIPS
MS-SSIM
NIQE
PSNR
SSIM
Results
Performance results of various models on this benchmark
Model Name | FED | FID | LLE | LPIPS | MS-SSIM | NIQE | PSNR | SSIM | Paper Title | Repository |
|---|---|---|---|---|---|---|---|---|---|---|
| WaveletCNN | 0.0964 | 16.472 | 2.702 | 0.2443 | 0.952 | 12.217 | 28.750 | 0.806 | Wavelet-SRNet: A Wavelet-Based CNN for Multi-Scale Face Super Resolution | - |
| UNET++ | - | - | - | - | 0.976 | - | 30.465 | 0.859 | - | - |
| SRFBN | 0.0984 | 20.032 | 2.066 | 0.2406 | 0.953 | 13.901 | 29.577 | 0.827 | Feedback Network for Image Super-Resolution | |
| Super-FAN | 0.1416 | 14.811 | 2.333 | 0.2357 | 0.913 | 8.719 | 25.463 | 0.729 | Super-FAN: Integrated facial landmark localization and super-resolution of real-world low resolution faces in arbitrary poses with GANs | - |
| HiFaceGAN | 0.0716 | 1.898 | 2.071 | 0.0723 | 0.971 | 6.961 | 30.824 | 0.838 | HiFaceGAN: Face Renovation via Collaborative Suppression and Replenishment | |
| SRGAN | 0.1097 | 4.396 | 2.269 | 0.1313 | 0.935 | 7.378 | 27.494 | 0.735 | Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network | |
| ESRGAN | 0.1107 | 3.503 | 2.261 | 0.1221 | 0.935 | 6.984 | 27.134 | 0.741 | ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks | |
| EDSR | 0.0843 | 20.605 | 2.003 | 0.2475 | 0.961 | 13.636 | 30.188 | 0.824 | Enhanced Deep Residual Networks for Single Image Super-Resolution |
0 of 8 row(s) selected.