add support for multiple labels #111
Conversation
caliban_toolbox/reshape_data.py
Outdated
| dimension_labels = ['fovs', 'stacks', 'crops', 'slices', 'rows', 'cols', 'channels'] | ||
| dimension_labels = ['fovs', 'stacks', 'crops', 'slices', 'rows', 'cols', 'compartments'] |
There was a problem hiding this comment.
This list of strings is used twice, maybe we can move it to the top as a global/constant:
DIMENSION_LABELS = ['fovs', 'stacks', 'crops', 'slices', 'rows', 'cols', 'compartments']
|
Is there a way to have project-wide global variables? I think the goal is to have consistent naming of the xarray dimensions across this whole repo, and have it match the nomenclature we're using in deepcell (hence the switch to compartments). Maybe if we define them as globals it will be more obvious that it's not some random hardcoded string? What do you think is the best way to handle that? |
|
Yes, I often like to define a |
|
Do you have an example somewhere of how you format that and like to import it? |
caliban_toolbox/settings.py
Outdated
| X_DIMENSION_LABELS = config('DIMENSION_LABELS', cast=list, | ||
| default=['fovs', 'stacks', 'crops', 'slices', | ||
| 'rows', 'cols', 'channels']) | ||
|
|
||
| Y_DIMENSION_LABELS = config('DIMENSION_LABELS', cast=list, | ||
| default=['fovs', 'stacks', 'crops', 'slices', | ||
| 'rows', 'cols', 'compartments']) |
There was a problem hiding this comment.
How does casting to a list work? Does it expect a comma-separated entry or something?
This PR adds the functionality to handle different types of segmentation labels.
It changes the nomenclature to match how deepcell-toolbox deals with the labels (compartments). It removes hardcoding of dimension names that was previously used, and takes the name directly from whatever array is supplied. It also propagates this through the pipeline.
This closes #68