Skip to content

smiler96/Image-Super-Resolution

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

45 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Image Super Resolution

Specifaication

  1. All the methods are executed on the nvidia 1066GPU.
  2. No pretrained on the 2x resolution dataset, 8x resolution training from scratch.
  3. No psnr calculation.

Example Results

Method Img1 Img2
LR
HR
EDSR_sigmoid
EDSR_None
RCAN
RCAN_InstanceNorm
RDN
AFN
DDBPN
DBPN-RES-MR
HAN-RES

Training Loss

Usage

Below are some examples showing how to run the main.py demos.

  • EDSR CVPR2017

$ >python main.py --phase 'train' --hr_train_path 'DIV2K_train_HR/' --lr_train_path 'DIV2K_train_LR_x8/' --hr_val_path 'DIV2K_valid_HR/' --lr_val_path 'DIV2K_valid_LR_x8/' --scale 8 --res_scale 0.1 --last_act 'sigmoid' --normalization 1 --augment

$ >python main.py --phase 'test' --test_file 'images/0801x8.png' --scale 8 --last_act 'sigmoid' --normalization 1

  • RCAN ECCV2018

$ >python main.py --phase 'train' --model 'RCAN' --hr_train_path 'DIV2K_train_HR/' --lr_train_path 'DIV2K_train_LR_x8/' --hr_val_path 'DIV2K_valid_HR/' --lr_val_path 'DIV2K_valid_LR_x8/' --scale 8 --n_rg 10 --n_rcab 20 --n_feats 64 --normalization 2 --augment

$ >python main.py --phase 'test' --model 'RCAN' --test_file 'images/0801x8.png' --scale 8 --n_rg 10 --n_rcab 20 --n_feats 64 --normalization 2

  • DDBPN CVPR2018

$ >python main.py --phase 'train' --model 'DDBPN' --hr_train_path 'DIV2K_train_HR/' --lr_train_path 'DIV2K_train_LR_x8/' --hr_val_path 'DIV2K_valid_HR/' --lr_val_path 'DIV2K_valid_LR_x8/' --scale 8 --n_feats 128 --nr 32 --n_depths 6 --normalization 2 --augment

$ >python main.py --phase 'test' --model 'DDBPN' --test_file 'images/0801x8.png' --scale 8 --n_feats 128 --nr 32 --n_depths 6 --normalization 2

  • DBPN-RES-MR Arxiv2019

$ >python main.py --phase 'train' --model 'DBPN_MR' --hr_train_path 'DIV2K_train_HR/' --lr_train_path 'DIV2K_train_LR_x8/' --hr_val_path 'DIV2K_valid_HR/' --lr_val_path 'DIV2K_valid_LR_x8/' --scale 8 --n_feats 256 --nr 64 --n_depths 7 --n_iters 3 --global_res --normalization 2 --augment

$ >python main.py --phase 'test' --model 'DBPN_MR' --test_file 'images/0801x8.png' --scale 8 --n_feats 256 --nr 64 --n_depths 7 --n_iters 3 --global_res --normalization 2

  • RDN CVPR2018

$ >python main.py --phase 'train' --model 'RDN' --hr_train_path 'DIV2K_train_HR/' --lr_train_path 'DIV2K_train_LR_x8/' --hr_val_path 'DIV2K_valid_HR/' --lr_val_path 'DIV2K_valid_LR_x8/' --scale 8 --D 20 --G 32 --C 6 --n_feats 64 --normalization 2 --augment

$ >python main.py --phase 'test' --model 'RDN' --test_file 'images/0801x8.png' --scale 8 --D 20 --G 32 --C 6 --n_feats 64 --normalization 2

  • AFN CVPR2020

$ >python main.py --phase 'train' --model 'AFN' --hr_train_path 'DIV2K_train_HR/' --lr_train_path 'DIV2K_train_LR_x8/' --hr_val_path 'DIV2K_valid_HR/' --lr_val_path 'DIV2K_valid_LR_x8/' --lr 1e-5 --batch_size 8 --scale 8 --n_feats 128 --n_l3 3 --act 'leaky_relu' --normalization 2 --augment

$ >python main.py --phase 'test' --model 'AFN' --test_file 'images/0801x8.png' --scale 8 --n_feats 128 --n_l3 3 --act 'leaky_relu' --normalization 2

  • HAN ECCV2020

    train explosion at 61 epoch when bs is 8

$ >python main.py --phase 'train' --model 'HAN' --hr_train_path 'DIV2K_train_HR/' --lr_train_path 'DIV2K_train_LR_x8/' --hr_val_path 'DIV2K_valid_HR/' --lr_val_path 'DIV2K_valid_LR_x8/' --scale 8 --n_rg 10 --n_rcab 20 --n_feats 128 --normalization 2 --augment

$ >python main.py --phase 'test' --model 'HAN' --test_file 'images/0801x8.png' --scale 8 --n_rg 10 --n_rcab 20 --n_feats 128 --normalization 2

About

Image-Super-Resolution

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages