When using ROIs in the torch_em Dataloader (I used: torch_em.default_segmentation_loader) there is a generic error message if the ROIs do not fit (e.g. empty or incorrect size).
Error message:
File "/home/freckmann15/miniforge3/envs/sam/lib/python3.11/site-packages/torch/utils/data/dataloader.py", line 631, in __next__
data = self._next_data()
^^^^^^^^^^^^^^^^^
File "/home/freckmann15/miniforge3/envs/sam/lib/python3.11/site-packages/torch/utils/data/dataloader.py", line 1319, in _next_data
raise StopIteration
StopIteration
This message is quite difficult to interpret and therefore it is difficult to fix.
The error can be fixed by having a min_shape for the ROIs so ROIs are at least the shape of the patch_shape used.
I would suggest a more specific error message for this case.
Description
When using ROIs in the torch_em Dataloader (I used:
torch_em.default_segmentation_loader) there is a generic error message if the ROIs do not fit (e.g. empty or incorrect size). Error message:This message is quite difficult to interpret and therefore it is difficult to fix. The error can be fixed by having a min_shape for the ROIs so ROIs are at least the shape of the patch_shape used. I would suggest a more specific error message for this case.