I have a particular case for FID calculation, where the user provides path to image directory. With the first approach this will involve some additional work and file preprocessing. To have the second approach, medigan guidelines will have to be updated, to require contributors to save each type of output data in separate directory. However, any additional sample use like statistics calculation will be very straightforward.
The first approach is already implemented in #36
What directory structure would be best for saved samples?
Currently I see two options:
First: Everything in one directory, that's simple to implement.
Second: Samples grouped by type.
I have a particular case for FID calculation, where the user provides path to image directory. With the first approach this will involve some additional work and file preprocessing. To have the second approach,
medigan
guidelines will have to be updated, to require contributors to save each type of output data in separate directory. However, any additional sample use like statistics calculation will be very straightforward. The first approach is already implemented in #36