ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks

The Super-Resolution Generative Adversarial Network (SRGAN) is a seminal work that is capable of generating realistic textures during single image super-resolution. However, the hallucinated details are often accompanied with unpleasant artifacts... (read more)

PDF Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Image Super-Resolution BSD100 - 4x upscaling SRGAN + Residual-in-Residual Dense Block PSNR 27.85 # 4
SSIM 0.7455 # 8
Image Super-Resolution FFHQ 256 x 256 - 4x upscaling ESRGAN FID 166.36 # 9
MS-SSIM 0.747 # 9
PSNR 15.43 # 11
SSIM 0.267 # 11
Face Hallucination FFHQ 512 x 512 - 16x upscaling ESRGAN FID 50.901 # 2
LPIPS 0.3928 # 2
NIQE 15.383 # 4
Image Super-Resolution FFHQ 512 x 512 - 4x upscaling ESRGAN PSNR 27.134 # 6
SSIM 0.741 # 5
MS-SSIM 0.935 # 5
LLE 2.261 # 4
FED 0.1107 # 6
FID 3.503 # 2
LPIPS 0.1221 # 2
NIQE 6.984 # 2
Image Super-Resolution Manga109 - 4x upscaling bicubic PSNR 24.89 # 21
SSIM 0.7866 # 21
Image Super-Resolution Manga109 - 4x upscaling SRGAN + Residual-in-Residual Dense Block PSNR 31.66 # 5
SSIM 0.9196 # 9
Image Super-Resolution PIRM-test ESRGAN NIQE 2.55 # 2
Image Super-Resolution Set14 - 4x upscaling SRGAN + Residual-in-Residual Dense Block PSNR 28.99 # 5
SSIM 0.7917 # 9
Image Super-Resolution Set5 - 4x upscaling SRGAN + Residual-in-Residual Dense Block PSNR 32.73 # 5
SSIM 0.9011 # 10
Image Super-Resolution Urban100 - 4x upscaling SRGAN + Residual-in-Residual Dense Block PSNR 27.03 # 8
SSIM 0.8153 # 7
Image Super-Resolution Urban100 - 4x upscaling bicubic PSNR 23.14 # 37
SSIM 0.6577 # 35

Results from Other Papers


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK SOURCE PAPER COMPARE
Image Super-Resolution FFHQ 1024 x 1024 - 4x upscaling ESRGAN FID 72.73 # 9
MS-SSIM 0.782 # 9
PSNR 19.84 # 9
SSIM 0.353 # 9

Methods used in the Paper


METHOD TYPE
Dropout
Regularization
Softmax
Output Functions
Max Pooling
Pooling Operations
PReLU
Activation Functions
Sigmoid Activation
Activation Functions
ReLU
Activation Functions
VGG
Convolutional Neural Networks
Residual Block
Skip Connection Blocks
Dense Connections
Feedforward Networks
Residual Connection
Skip Connections
Leaky ReLU
Activation Functions
PixelShuffle
Miscellaneous Components
VGG Loss
Loss Functions
SRGAN Residual Block
Skip Connection Blocks
SRGAN
Generative Adversarial Networks
Relativistic GAN
Generative Adversarial Networks
Convolution
Convolutions
Batch Normalization
Normalization
GAN
Generative Models