This plugin provides a wrapper around Cellpose's deep learning segmentation models. Current features:
Options for most of the basic parameters from the Cellpose API, including mode-switching between detection of nuclei and cells with the inbuilt models.
Supports using a custom model file, in case you've re-trained the stock models or made your own.
I've provided the option to run the model on a GPU if it's available. This has additional dependencies and won't work on all systems, so there's also a 'GPU test' button
Works in 2D or 3D. 3D mode on a GPU requires a lot more VRAM so users may want to resize their images down before trying to run.
Users can optionally save the probability image alongside the detected objects.
Current interface:
Current figure display (with probability export on):
I'm not sure how best to setup tests/integration into the rest of the repo. pip install cellpose gets you running on a CPU, but will add pytorch and other libraries as dependencies. Since full GPU setup is a bit complex it might be worth moving towards setting requirements per-plugin rather than across the whole repository?
This plugin provides a wrapper around Cellpose's deep learning segmentation models. Current features:
Current interface:
Current figure display (with probability export on):
I'm not sure how best to setup tests/integration into the rest of the repo.
pip install cellpose
gets you running on a CPU, but will add pytorch and other libraries as dependencies. Since full GPU setup is a bit complex it might be worth moving towards setting requirements per-plugin rather than across the whole repository?