`create_dataset.py` script creates patches with size 0

[nilsleh]

Hi @paren8esis , I have downloaded the data from dropbox and tried to run the conversion script create_dataset.py as instructed in the README. However, there are some cases where patches of size 0 are created. The culprit in those cases is that the label is larger than the other data modalities, for example:

label (1, 9538, 12432)
mod_500_cloud_post (1, 6218, 12432)
mod_500_cloud_pre (1, 6218, 12432)
mod_500_post (7, 6218, 12432)
mod_500_pre (7, 6218, 12432)
sea_mask (1, 6218, 12432)
sen2_10_post (4, 6218, 12432)
sen2_10_pre (4, 6218, 12432)

this is for <HDF5 group "/2020/1" (9 members)> (PosixPath('floga/FLOGA_dataset_2020_sen2_10_mod_500.h5'))

Is there perhaps some data misalignment issue?

[nilsleh]

@paren8esis sorry to ping again, do you happen to have any insights on this?