zhangyi-3 / KBNet

KBNet: Kernel Basis Network for Image Restoration
MIT License
199 stars 25 forks source link

The SSIM of the SIDD does not match #9

Closed Cynicarlos closed 8 months ago

Cynicarlos commented 11 months ago

Dear author: As I show here :

results_paper

result_test

The PSNR matches with the relults in your paper, but the SSIM does not. What is wrong?

Thank you!

jacky22043 commented 9 months ago

Hello, I get the same results in paper. Here is my setting: denoise_SIDD

  1. Download SIDD validation data, run python download_data.py --noise real --data test --dataset SIDD
  2. Download the Denoise folder pre-trained models. 2.1 Replace all files in denoise to the location of KBNet-main\Denoising.
  3. And make sure copy the basicsr folder under root folder to Denoising folder.
whhwhhhh commented 8 months ago

Where to download sidd.pth,I can't find it 650

Cynicarlos commented 8 months ago

Where to download sidd.pth,I can't find it 650

image

whhwhhhh commented 8 months ago

thanks

---Original--- From: @.> Date: Sun, Mar 10, 2024 12:33 PM To: @.>; Cc: @.**@.>; Subject: Re: [zhangyi-3/KBNet] The SSIM of the SIDD does not match (Issue #9)

Where to download sidd.pth,I can't find it

image.png (view on web)

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

whhwhhhh commented 8 months ago

OOPS。。。。。I still can't find this ,can you share me zhe link or url?

Cynicarlos commented 8 months ago

OOPS。。。。。I still can't find this ,can you share me zhe link or url?

https://mycuhk-my.sharepoint.com/personal/1155135732_link_cuhk_edu_hk/_layouts/15/onedrive.aspx?ga=1&id=%2Fpersonal%2F1155135732%5Flink%5Fcuhk%5Fedu%5Fhk%2FDocuments%2Fshare%2FKBNet%2FDenoising%2Fpretrained%5Fmodels

whhwhhhh commented 8 months ago

652 eemmmmm,maybe something is wrong

Cynicarlos commented 8 months ago

652 eemmmmm,maybe something is wrong

The author gave the link in https://github.com/zhangyi-3/KBNet/blob/main/Denoising/README.md#Real-Image-Denoising you can find all models there