Residual Dense Network for Image Super-Resolution

A very deep convolutional neural network (CNN) has recently achieved great success for image super-resolution (SR) and offered hierarchical features as well. However, most deep CNN based SR models do not make full use of the hierarchical features from the original low-resolution (LR) images, thereby achieving relatively-low performance... (read more)

PDF Abstract CVPR 2018 PDF CVPR 2018 Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Image Super-Resolution BSD100 - 4x upscaling RDN PSNR 27.72 # 13
SSIM 0.7419 # 17
Color Image Denoising CBSD68 sigma50 Residual Dense Network + PSNR 28.34 # 1
Image Super-Resolution Manga109 - 4x upscaling RDN PSNR 31.00 # 15
SSIM 0.9151 # 13
Image Super-Resolution Set14 - 4x upscaling RDN PSNR 28.81 # 15
SSIM 0.7871 # 18
Image Super-Resolution Set5 - 4x upscaling RDN PSNR 32.47 # 14
SSIM 0.8990 # 15
Image Super-Resolution Urban100 - 4x upscaling RDN PSNR 26.61 # 15
SSIM 0.8028 # 14

Methods used in the Paper


METHOD TYPE
Convolution
Convolutions
Concatenated Skip Connection
Skip Connections
Batch Normalization
Normalization
ReLU
Activation Functions
Dense Block
Image Model Blocks