RDN-TensorFlow icon indicating copy to clipboard operation
RDN-TensorFlow copied to clipboard

Set5 test set evaluation result

Open LiUzHiAn opened this issue 6 years ago • 6 comments

Hi, I used the model you pretrained with 200 epochs and evaluated the model on Set5 test set. But I only got ~25.5 average PSNR which is somewhat poor. Anybody same with me? Should I train the model continuely ?

LiUzHiAn avatar Feb 25 '19 10:02 LiUzHiAn

hello!How long have you been training this network?

13135675192 avatar Mar 05 '19 01:03 13135675192

I used Set91 and BSD200 as training set. Traning data 200 epochs on Colab nearly cost 5 hrs

LiUzHiAn avatar Mar 05 '19 01:03 LiUzHiAn

I used DIV2K as training set. and set "matlab -true "on GTX-1080ti nearly cost 8days. The result is 34.5average PSNR,the testing set is set5

13135675192 avatar Mar 05 '19 02:03 13135675192

oh, the result is not that good even worse than SRCNN. Have you got any improvement ?

LiUzHiAn avatar Mar 11 '19 06:03 LiUzHiAn

@13135675192 Hey, how many training images do you have. After preprocessing the traing images, I got about 500 thousand images. When I train this network with scale of 3 for seven days , the trainning step is just 800 thousand.

sxd0071 avatar Mar 11 '19 11:03 sxd0071

@ 13135675192 May I ask you 34.5 average PSNR at scale3 or 2?

Can you use your checkpoint folder for reference? My mailbox is [email protected]

kong1060553768 avatar Jun 19 '19 02:06 kong1060553768