LoSealL / VideoSuperResolution

A collection of state-of-the-art video or single-image super-resolution architectures, reimplemented in tensorflow.
MIT License
1.61k stars 295 forks source link

frame is empty. #93

Closed wuwei6 closed 3 years ago

wuwei6 commented 4 years ago

Following the guide here https://github.com/LoSealL/VideoSuperResolution/blob/master/Data/README.md

Assign your own database URL.

Sorry, I seem to have accidentally deleted the last question. I can't open your dataset URL in my area, So I don't know how to assign my own database URL. May I ask what should I do if I want to train and test off-line?

LoSealL commented 4 years ago

You misunderstood.

If you have already prepared your off-line dataset, what you need to do is:

blackhumor22 commented 4 years ago

Hello, I have the same problem, I have followed the path of dataset.yml to add. The problem is still not solved, can you give me an example? For example the complete path of set5 data, thank you.

wuwei6 commented 4 years ago

Thank you for your help! I use the DRRN and the VDSR model for training. I got the test results by using the set5 dataset, but it is only 0.03 higher than the PSNR calculated by bicubic interpolation. MAE is about 120. I added the epoch, but it didn't change. Should I change the parameters?

LoSealL commented 3 years ago

Closed for a long time no response. Reopen if you needed.