Releases: vanvalenlab/deepcell-applications
0.4.1
The main feature of release v0.4.1 is the addition of imagecodecs
to support processing of compressed OME tiff files.
🧰 Maintenance
Bump dependencies to support compressed OME TIFs @rossbar (#36)
Add imagecodecs
to requirements.
Also includes a temporary pin for scikit-image to avoid issues with the latest release.
General maintenance of test infrastructure @rossbar (#34)
Similar to vanvalenlab/deepcell-tf#639, minus the explicit add for Python 3.10 support which will need to wait for another deepcell-tf release.
Update default actions to latest version. @rossbar (#33)
Rm deprecation warnings in action logs. See also: vanvalenlab/deepcell-tf#653
0.4.0
This release includes the updated Mesmer model that was released in the most recent deepcell-tf build. It provides better nuclear segmentation in areas without true cells. It also has a slight tweak to the post-processing for cell segmentation to better balance between segmentation errors and inclusion of cell area.
Pre-release updates @ngreenwald (#29)
Update to new Mesmer model @ngreenwald (#28)
0.3.1
🐛 Bug Fixes
Add "both" to the argument parser. @willgraf (#25)
Adds "both" as a valid value for --compartment
to better match the model and removes "membrane".
Fixes #24
🧰 Maintenance
Adding testing for argparse @ngreenwald (#22)
We currently don't test the argparse functionality in the repo. This PR moves the argparse function to utils and adds testing to make sure parameters are passed appropriately, and that errors are raised when required.
0.3.0
0.2.1
🐛 Bug Fixes
Add `--squeeze` to all the child subparsers. @willgraf (#20)
--squeeze
was added as an optional argument in #17, but it was added to the main argument parser instead of the parent. This means that the mesmer
command does not recognize it.
This PR resolves the issue by adding it to the parent
argparser instead of the primary one. This allows mesmer
and all other subparsers to inherit the --squeeze
arg.
0.2.0
🚀 Features
Add `--squeeze` to CLI arguments. @willgraf (#17)
If passed, call np.squeeze
on the output tensor before saving as a tiff.
Closes #16
🐛 Bug Fixes
🧰 Maintenance
Upgrade DeepCell to version 0.10.0 @willgraf (#18)
The latest release of DeepCell (0.10.0) fixes several performance issues including:
- improved batching in the
Applications
- Support individual parameter passing to Mesmer application
- Allow preprocessing_fn and postprocessing_fn to be overridden.
- Update Tensorflow to 2.5.1 (fixes several vulnerabilities).
0.1.0
0.1.0
Features
- New
deepcell_applications
module with fully tested application helper functions - Support multiple channels from a single channel via
nuclear-channel
andmembrane-channel
- Enable multiple applications via
argparse
subparsers and Application class name - Add GitHub Actions workflows for testing, docker image builds, and release drafts
Bugfixes
- Pin
deepcell~=0.9.2
and fixget_image
imports