Unused global variables in cellfinder detection?

[alessandrofelder]

IIUC (and I may well not) from inspecting the detection code, the global variables in

NETWORK_VOXEL_SIZES = [5, 1, 1]
CUBE_WIDTH = 50
CUBE_HEIGHT = 20
CUBE_DEPTH = 20

are partially unused: I can only find references to NETWORK_VOXEL_SIZES[0] and CUBE_DEPTH.
Should the unused variables be removed?

There is similar code in train_yaml (but CUBE_HEIGHT is 50, confusingly) which could be refactored and better documented as part of this issue, too?

[adamltyson]

I'm sure they used to be used in a few places, but this may have got lost in the various re-writes. Not sure if we should get rid of them, or make sure they're used throughout (and make sure they're correct!).

[Gautghadi-1605]

Hi, I’m exploring cellfinder as part of my GSoC 2026 preparation, and I’d like to investigate this issue if that sounds okay. I’ll first trace where these constants are currently used, and then decide whether the right fix is removal or refactoring. If there’s a preferred scope for a first PR, please let me know.

[Gautghadi-1605]

I checked the current references and it looks like CUBE_WIDTH and CUBE_HEIGHT in cellfinder/napari/detect/detect.py are not used, while the related constants in train_yaml.py are still used. I’m planning a small PR to clean up the unused constants in the detection file first.

[Gautghadi-1605]

I’ve opened a PR for this cleanup: #604