NBNet icon indicating copy to clipboard operation
NBNet copied to clipboard

NBNet: Noise Basis Learning for Image Denoising with Subspace Projection

Results 13 NBNet issues
Sort by recently updated
recently updated
newest added

when run ‘python test.py -d ./data -c ./NBNet_mge.pkl’,occur error ‘FileNotFoundError: [Errno 2] No such file or directory: './data/ValidationNoisyBlocksSrgb.mat'’,does the test dataset need a .mat file?

Thank you for your work. NBNet requires size 128 as its input, but when testing img size > 128, if we clip the input to size 128 and then merge...

Hello, how many pictures did you use in your training set

![UJP9DFD_NJ~U7JD8$GB _(3](https://user-images.githubusercontent.com/41667602/159398175-4d9c7e50-eff6-4f94-8db2-c81bd7d7c848.png) Why does this step get stuck before the test loads the model?Unable to run further down。 ![image](https://user-images.githubusercontent.com/41667602/159398415-1adf74b9-df9d-4e90-ae1e-3ad8414cab64.png)

![image](https://user-images.githubusercontent.com/73509952/119304476-de44c880-bc99-11eb-9a57-1f90e8324b4c.png)

It is a strange issue that the loss will blow up to Nan in linux ubuntu 20.04 after I install the cuda, without cuda the loss decrease normally. environment: megengine...

Would you release the pre-trained checkpoint for AWGN removal? I want to evaluate the NBNet on some other dataset, such as Urban100. @Sumching

![image](https://user-images.githubusercontent.com/18656390/134484668-d1e0346a-0f4b-4554-8dbb-9253e3b22598.png)

Is the training set full data, medium or small data of SIDD data set used?