search

juglab / n2v

This is the implementation of Noise2Void training.
Other
385 stars 107 forks source link

Config should not need X #139

Closed msseibel closed 10 months ago

msseibel commented 1 year ago

I am writing an enhancement such that you don't need to pass the whole training data to N2VConfig but rather the properties which are necessary such as the number of samples, number of channels, mean and standard deviation and so on.

Ultimately, I would like to read the images only into RAM if they are need for training.

Do you generally accept pull requests?

jdeschamps commented 1 year ago

Hi!

Yes, you are entirely right! It would be great to have a lazy data loader... We are very happy with PRs!! We will try to review it asap, just know that we are a bit stretched at the moment.

Now, a disclaimer: we are not actively developing this repository. There was a recent update following a publication, but other than that we only try to answer issues and fix obvious bugs. The reason for that is that we are working on a new library (which includes N2V) that we hope to release this year. This library will include lazy data loaders and many other features.

jdeschamps commented 10 months ago

I will close this issue now.