Shaosifan / HSENet

Hybrid-Scale Self-Similarity Exploitation for Remote Sensing Image Super-Resolution (accepted by TGRS)
39 stars 5 forks source link

can not train the model as instructed in readme file #5

Closed jhrcws closed 1 year ago

jhrcws commented 1 year ago

as your side in the "train" part:# x4 python demo_train.py --model=HSENET --dataset=UCMerced --scale=4 --patch_size=192 --ext=img --save=HSENETx4_UCMerced

x3

python demo_train.py --model=HSENET --dataset=UCMerced --scale=3 --patch_size=144 --ext=img --save=HSENETx3_UCMerced

x2

python demo_train.py --model=HSENET --dataset=UCMerced --scale=2 --patch_size=96 --ext=img --save=HSENETx2_UCMerced

I can not find the model, dataset and scale setting parameters in demo_train.py file 1673098212814 1673098250631

Shaosifan commented 1 year ago

Hi, the setting parameters of model, dataset, and scale can be found in option.py.

jhrcws commented 1 year ago

thanks for your reply. At the same time, I see many modules and .py files in the HESNet-main files, and as a newbie in deep leaning field, I am very confused what to do to reproduce the results in your paper. So would you tell me what should I do in the first sencond third... steps to reproduce the results in the paper? I will appreciate for your valuable help.

a23b6971cc6a8c8d61e9de0e23288ab

Shaosifan commented 1 year ago

The main steps to use this repository can be found in README.md

  1. Install some requirements.
  2. Download the dataset and set paths in data/init.py
  3. Train HSENet using demo_train.py.
  4. Test the trained HSENet using demo_deploy.py.
  5. Make evaluations using calculate_PSNR_SSIM.py.