0.17.2

🧰 Maintenance

Bump deepcell version and add tracking parameters @msschwartz21 (#199)

This PR preps the consumers for the tracking paper. It bumps up the versions of deepcell packages and adds additional tracking parameters that have been recently exposed.

This update has been tested in a test deployment of the kiosk, but I would like to hold off on merging it until we have resolved vanvalenlab/deepcell-tf#665 and released a new version of deepcell with the fix.

0.17.1

🐛 Bug Fixes

Save and expose sorted output names @elaubsch (#196)

This PR fixes this issue. The GRPC client returns a dictionary with keys corresponding to the output names. These names were previously sorted but the sorted keys were not saved. Therefore, when these keys were exposed and used, sometimes they were in the wrong order.

0.17.0

🧰 Maintenance

Update to Tensorflow 2.8 @elaubsch (#195)

This PR updates the TF version run by the kiosk, including the following changes:

• Bumping the required TF version to 2.8
• Bumping the required version of deepcell packages to those running TF 2.8
• Adding support for Python 3.9

Breaking Changes

• Drop support for Python 3.6

0.16.2

🐛 Bug Fixes

Add imagecodcs to requirements @elaubsch (#193)

This change fixes #192

0.16.1

🐛 Bug Fixes

Expose GRPC model output names @elaubsch (#191)

This PR makes changes to expose the output names of the GRPC-wrapped model. These changes allow the outputs of the TF-served model to be formatted into a dictionary. They are related to this spots PR.

🧰 Maintenance

Cache pip environment for Github actions @elaubsch (#190)

This PR introduces a change to cache the pip environment created to run the test Github action. With this change the tests run in under 2 minutes instead of 7-8 minutes.

0.16.0

🚀 Features

Add multi-batch prediction for the segmentation and Polaris consumers @elaubsch (#189)

This PR adds support for multi-batch images for the segmentation and Polaris consumers. The segmentation consumer receives a variable dimension_order from the frontend (see this PR). This variable has been added to the data generated by the Polaris consumer for segmentation jobs.

We have also added more comprehensive testing of the segmentation consumer for input images of different dimension orders. This involved changes to the DummyStorage object to allow the dimensions of the images it generates to be specified.

We have removed a line from the get_image utils function that expands the dimensions of loaded TIFF files by default, because this behavior complicated the handling of images with dimensions BXYC or CXYB. This change had little impact on the other functions that call get_image, because this channel was often squeezed out immediately after loading.

0.15.1

🐛 Bug Fixes

Update assignment for channels variable in Caliban consumer @elaubsch (#188)

This PR updates the channels value for segmentation hashes created by the Caliban consumer. The changed format in this PR.

0.15.0

🚀 Features

Polaris super consumer @elaubsch (#183)

This PR introduces a Polaris "super consumer" that handles spot detection and cell segmentation for singleplex FISH analysis.

• The consumer that handles spot detection tasks - previously called the PolarisConsumer - has been refactored to the SpotDetectionConsumer.
• The label detection model has been removed from the segmentation consumer and the channel handling for nuclear and whole-cell segmentation is now done manually.
• The MAX_SCALE variable has been increased from 3 to 10 to allow for segmentation of 100x images, which are typical for FISH experiments.

0.14.0

🚀 Features

Adding Polaris consumer @elaubsch (#181)

Adding Polaris consumer and tests

🧰 Maintenance

Update copyright to 2022 @msschwartz21 (#182)

0.13.0

🚀 Features

Refactor TrackingConsumer to CalibanConsumer. @willgraf (#180)

Similar to #160, we are renaming the tracking workflow from Tracking to Caliban. The old options are still supported (tracking consumer type, TRACKING_MODEL, etc) but have been deprecated in favor of the caliban versions.

Additionally, the drift correction option has been removed, as it has never worked and relies on deprecated skimage code. Removes scikit-image from the dependencies, as the drift correction is the only place where it is used.

🐛 Bug Fixes

Verify image size before loading the image. @willgraf (#179)

Fixes #176

🧰 Maintenance

Update dependencies and upgrade Docker base image. @willgraf (#178)

• Update base image to Python 3.9-slim-buster.
• Add tifffile to requirements and remove scikit-image. The project depends on tifffile which is no longer included in skimage versions that are supported by newer Python versions.
• Update a few dependencies to their latest patches.

Upgrade `deepcell` to 0.11.x @willgraf (#177)

Also updates deepcell-tracking to 0.5.2