This is Homework4 Project for NYCU VRDL class I trained MSRResNet and upscale testing images with scaling factor 3x
PSNR : 28.1258
To run the project
- Install pytorch version 10.2 (earlier versions may also work)
- Install dependencies
make install
make download_data
The data will be automatically downloaded and unzipped in VRDL_HW4/data
make reproduce
The script will automatic download trained model checkpoints from google drive and run evaluation on testing dataset.
The resulting images will be stored at VRDL_HW4/KAIR_Repo/superresolution/msrresnet_psnr/images/evaluation{iterations}/
make train
The training result will be stored at VRDL_HW4/KAIR_Repo/superresolution/msrresnet_psnr/