MCCNN
MCCNN copied to clipboard
viscom-ulm
Path to cuda
Hello, I'm a student and I'm interested in your code. I tried to compile tensorflow operations by following your instructions, but I get an error. The server that I'm working on has different versions of cuda installed but when I enter whereiscuda on the terminal it shows "usr/local/cuda" so I'm using this path. Is this correct? Or should the path be "usr/local/cuda-11.2" since tensorflow works with that specific version?
Hello, thanks for your interest in our code. This code is quite old and I did not try it with this version of Cuda. But it seems that you are doing everything correctly. However, I would recommend you to use a newer implementation of this work in Pytorch that you can find in the following repository:
https://github.com/phermosilla/pccnn_weightinit_experiments
I am still working on the tutorials, etc. But it should be updated in the following weeks before ECCV at the end of October.
Does this new code works the same as MCCNN Library? Because I'm trying to use this code https://github.com/phermosilla/TotalDenoising and it needs MCCNN.
No, sorry. You will need to use this library for the code in total denoising. I recommend then using a docker container where everything is setup properly.