esw0116/Clustering_SwinIR

Code optimization target

models/network_onlyattnnoir_gumbel_compensate2_simple2.py

Measure test result / time of ICASR and SwinIR

Install Pytorch and other stuffs (please refer to requirements.txt).

Get DIV2K validation dataset for measuring the runtime

• HR: http://data.vision.ee.ethz.ch/cvl/DIV2K/DIV2K_valid_HR.zip
• LR: http://data.vision.ee.ethz.ch/cvl/DIV2K/DIV2K_valid_LR_bicubic_X2.zip

Download Pretrained models to report PSNR and SSIM

• SwinIR: https://drive.google.com/drive/folders/12UrZAg06T7l_ElqLYAtzKGc7b81Ldh-_?usp=sharing
• ICASR(Ours): https://drive.google.com/drive/folders/1Cf3wExjvVtA_bIktI8RU3C8eGfEYgUoy?usp=sharing

Don't need to download pretrained models if not measuring PSNR, SSIM

Change dataset path in 'main_test_clustertransformer.py' (Line 35-49)

Move to file 'clustertest.sh'

• Use Line 2 to test ICASR (Simplified version)
• Erase '--model_path ~~~' if only need to check runtime.
• add '--bs ' for increasing batch size, '--cpu' for cpu testing

Run code by 'sh clustertest.sh'