kevinjohncutler / omnipose

Omnipose: a high-precision solution for morphology-independent cell segmentation
https://omnipose.readthedocs.io
Other
85 stars 29 forks source link

Example1 notebook not working #8

Closed cnstt closed 1 year ago

cnstt commented 2 years ago

When executing the example1 notebook, I discovered several "issues":

For the first 2 I could open a PR to correct it ;)

cnstt commented 2 years ago

For the import error I encountered just changing it to files = io.get_image_files(basedir, "") works :)

kevinjohncutler commented 2 years ago

Thanks @cnstt! I'll make these changes. Some of the functions have been updated and that's the source of these errors. I'll have to do some more digging for the model download error.

alix-pham commented 1 year ago

Hi!

I'm trying to use Omnipose for bacterial segmentation and the last issue mentioned here is quite problematic for me. Is there one of the current models that can replace the 'bact_phase_omni' model? ('cyto', 'nuclei', 'tissuenet', 'livecell', 'cyto2', 'general', 'CP', 'CPx', 'TN1', 'TN2', 'TN3', 'LC1', 'LC2', 'LC3', 'LC4')

EDIT: I tried with 'general' but then, most of the keyword arguments do not exist anymore in cellpose.models.eval—I would guess because everything moved to the omnipose repo? I can see them (OMNI_MODELS) in core.py but I don't understand how to use the different functions to generate the masks (compute_masks? get_masks?)

Or is there somewhere the versions of cellpose and/or omnipose that were using that model (the one used in the Omnipose paper, I would guess)

Thank you very much in advance, Best,

Alix

kevinjohncutler commented 1 year ago

Sorry I didn't post here that those tasks have been completed for some time, thanks @cnstt. @alix-pham, those models are part of Cellpose2.0 and I have not yet made them available in Omnipose. This should be as simple as updating the name list in my Cellpose fork. You should be able to see bact_phase_omni, and if not, that means you do not have the latest version of Omnipose installed (use pip install git+...). I will eventually get the PyPi release updated, it just means that all my forked dependencies also need to be released on PyPi - and that's just been too much to juggle. Git installs don't care if dependencies are on GitHub or PyPi.