wyf0912 / LLFlow

[AAAI 2022] The code release of paper "AAAI Low-Light Image Enhancement with Normalizing Flow"
Other
292 stars 36 forks source link

Result mismatch #26

Closed zhanggengchen closed 1 year ago

zhanggengchen commented 2 years ago

Hello, I clone the code and test it on LOL dataset (eval15). I only modify the paths to dataset and pre-trained model in confs/LOL-pc.yml and run

python test.py --opt your_config_path

I got following results (PSNR 25.00) which is slightly different with the results in paper (PSNR 25.19)

截屏2022-08-22 23 58 09

I would like to know what causes the problem.

McFly-Jia commented 2 years ago

I also had the same problem.

wyf0912 commented 1 year ago

I retrain a model when I prepare the release of the code. This may cause a slight performance difference.

I will upload a new model soon.

wyf0912 commented 1 year ago

You can achieve almost the same result by changing the line in test.py from mean_out = cv2.cvtColor(sr_t.squeeze().permute(1,2,0).cpu().numpy(), cv2.COLOR_BGR2GRAY).mean() to mean_out = sr_t.view(sr_t.shape[0],-1).mean(dim=1)

image
zhanggengchen commented 1 year ago

Thanks for your reply, it's helpful for me 😃