Can't train a model with only 4 images

[thibaultdvx]

clinicadl_handbook/src/training_classification.py

Line 316 in caab842

!clinicadl train classification data_oasis/CAPS_example slice_classification_t1 data_oasis/split/4_fold/ data_oasis/maps_classification_2D_slice_resnet18 --n_splits 4 --architecture resnet18

There are only 4 images in the OASIS example dataset. It is not enough to train a model, as test and validation tests must contain at least 2 images (1 AD and 1 CN).

[thibaultdvx]

@NicolasGensollen I would like to add a file with more data (10 images, just enough to train a model) on the server. How can I do that?

[NicolasGensollen]

@thibaultdvx You need administrative rights on the machine to do that.
I think the easiest way would be that you put them on the Lustre such that I can get them and upload them.
WDYT ?

[thibaultdvx]

@NicolasGensollen sounds good to me. I'll let you know when it is done.