You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The connect_four function in structure_detection.py is called by structure_splitting.py:ball_filter_imgs with planes of type uint16 . Inside the connect_four function, we assign plane[x,y] to struct_id, a uint64. If there are more than 2**16-1 (temporary struct id) this will lead to confusion between structure with id 1 and structure 2**16!
To Reproduce
Pass cellfinder.detect a dataset that contains more than 2**16 individual structures.
Expected behaviour
Cell detection should be able to deal with >65000 structures.
Computer used (please complete the following information):
22.04.1-Ubuntu on a Dell Desktop
The text was updated successfully, but these errors were encountered:
For completeness sake (so we know what happened), this used to work in Cython, because plane (then layer) used to consist of uint64 s... so this was introduced as part of the numba refactoring
Describe the bug
The
connect_four
function instructure_detection.py
is called bystructure_splitting.py:ball_filter_imgs
with planes of typeuint16
. Inside the
connect_four
function, we assignplane[x,y]
tostruct_id
, auint64
. If there are more than2**16-1
(temporary struct id) this will lead to confusion between structure with id1
and structure2**16
!To Reproduce
Pass cellfinder.detect a dataset that contains more than 2**16 individual structures.
Expected behaviour
Cell detection should be able to deal with >65000 structures.
Computer used (please complete the following information):
22.04.1-Ubuntu on a Dell Desktop
The text was updated successfully, but these errors were encountered: