search

Zheng222 / IMDN

Lightweight Image Super-Resolution with Information Multi-distillation Network (ACM MM 2019, Winner Award of ICCVW AIM 2019 Constrained SR Track1&Track2)
429 stars 69 forks source link

PSNR is slightly lower in my experiments #16

Closed JingyunLiang closed 4 years ago

JingyunLiang commented 4 years ago

I tested IMDN_X4 using the provided checkpoint and Set5 dataset. The PSNR is 32.19. I also trained the IMDN_X4 from scratch strictly following your code. The best PSNR is 32.14. These two results are slightly lower than the reported 32.21 in Table 5.

What's the possible reason for the two differences? Do I need to train a model based on your pre-trained checkpoint? (i.e., --pretrained checkpoints/IMDN_x4.pth) If so, how was this checkpoint trained?

Lininggggggg commented 4 years ago

i have the same issus about this , could you tell me? dear author

Lininggggggg commented 4 years ago

what more, why the test time is different so large with different environment ? how should i to choose the best environment for my workplace? https://github.com/Zheng222/IMDN/issues/4#issue-506186607 and i found this link but i still have trouble about this? could you tell me? Thanks!!!

Zheng222 commented 4 years ago

@JingyunLiang Please use MATLAB to test PSNR values. If you want to further improve the PSNR, please train more steps when the learning rate is 2e-4.

Zheng222 commented 4 years ago

@Lininggggggg This problem does not occur on my machine. You can test time using KAIR and then compare it with our testing time.

Lininggggggg commented 4 years ago

OK,thank you very much!!!

JingyunLiang commented 4 years ago

Thank u.