ilastik 1.4.1b15

Summary

• Future-proofing and backwards compatibility of the Precomputed implementation
• UX improvements in data selection and object classification
• Default export in compressed hdf5

What's Changed

• Object size filter: Increase permitted value range by @btbest in #2784
• fixed skimage.segmentation.slic signature change by @k-dominik in #2792
• Prettify dataset table by @btbest in #2793
• Labeling GUI: Copy colortable by @btbest in #2786
• remove OpBigTiffReader by @k-dominik in #2799
• better "debug" for error in command line usage by @k-dominik in #2763
• Reimplement urldatasetinfo as legacy by @btbest in #2805
• Use key-strings to identify multiscales by @btbest in #2801
• Change default export format to compressed hdf5 by @k-dominik in #2812
• Multiscale: Support project files containing an integer working_scale by @btbest in #2816
• Data Export: Update buttons after removing lane by @btbest in #2815
• Fixed OpMultiChannelSelector by @k-dominik in #2748
• Ensure all object features get exported to table by @k-dominik in #2814
• Data selection: Remove inline Add buttons immediately by @btbest in #2821
• Added note that data is copied to project file by @k-dominik in #2818
• Improve web loading gui by @btbest in #2820
• Multiscale: Save scale by @btbest in #2826
• ci: circumvent disk space limitation on D: by @k-dominik in #2817
• let pytest use pytest.ini to discover tests by @k-dominik in #2745
• Fix pytest8 by @btbest in #2808
• Tests: Deflake scale combobox in test on MacOS by @btbest in #2806
• Tests: Mock user interaction in ImageFileDialog by @btbest in #2819
• Tests: Fix mock table and assert combobox attempts to change by @btbest in #2804

Full Changelog: 1.4.1b13...1.4.1b15

ilastik 1.4.1b13

Summary

• Simple multiscale support for remote datasets in Neuroglancer Precomputed format.

What's Changed

• lazyflow: use exception chaining in requests by @k-dominik in #2760
• Remove obsolete OpTransposeSlots by @btbest in #2762
• OpDataSelection: Default file exports to project file path by @btbest in #2774
• bump numpy 1.22 by @k-dominik in #2775
• try larger resource class for tests on circleci by @k-dominik in #2767
• Bump conda-incubator/setup-miniconda from 2 to 3 by @dependabot in #2780
• Multiscale v1 by @btbest in #2782

Full Changelog: 1.4.1b10...1.4.1b13

ilastik 1.4.1b10

fixed incompatible xarray versions that led to failures in the Neural Network workflow.

1.4.1b9

Summary

• fix neural network workflow with updated zenodo api

What's Changed

• OpDataSelection: Directly obtain data-reader by @btbest in #2757
• Remove string concatenation when raising by @btbest in #2759
• bump tiktorch by @k-dominik in #2765
• exclude broken xarray version in conda pkg by @k-dominik in #2768

Full Changelog: 1.4.1b7...1.4.1b9

1.4.1b7

What's Changed

• fix test mouse move by @k-dominik in #2736
• make NN-related operators aware of tensor ids by @k-dominik in #2732
• replace vigra for tiff reading with tifffile by @k-dominik in #2733
• bigger shape in 'z' for tiffs means faster browsing by @k-dominik in #2737
• Fix devenv by @k-dominik in #2726
• Disabled addRemoveData test by @k-dominik in #2742
• Undo elf fix by @k-dominik in #2734
• Data input: Phrase axistags error message more user-friendly by @btbest in #2747
• First try most compatible H5 saving, otherwise use v1.8 by @oanegros in #2746
• reduce number of deprecation warnings by @k-dominik in #2740
• Request object features per plugin in parallel by @k-dominik in #2750
• experimental API refactor by @k-dominik in #2752

Experimental API changes

the generic ilastik.experimental.api.from_project_file function is now deprecated. Please use ilastik.experimental.api.PixelClassificationPipeline.from_ilp_file instead. New workflows will be made available only in the explicit fashion in the future.
Also the PixelClassificationPipeline.predict method was deprecated in favor of a more explicit PixelClassificationPipeline.get_probabilities method. Idea is to add get_OUTPUT methods to existing/new piplelines that match the headless export names.

New Contributors

• @oanegros made their first contribution in #2746

Full Changelog: 1.4.1b6...1.4.1b7

1.4.1b6

What's Changed

• Restyle label table by @btbest in #2710
• bump python version to 3.9 by @k-dominik in #2703
• Support for Apple Silicon (Osx on arm 64) by @k-dominik in #2709
• increase timeout to launch tiktorch by @k-dominik in #2720
• Fix conservation tracking tests by @btbest in #2716
• unfreeze cache before exporting data from tda workflow by @k-dominik in #2723
• hacky fix for elf limiting threads by @k-dominik in #2724

Full Changelog: 1.4.1b1...1.4.1b6

1.4.1b1

What's Changed

• fix elf ws now returns uint64 by @k-dominik in #2690
• mnt: update pytorch versions by @k-dominik in #2692
• mnt: wsdt package not used by @k-dominik in #2685
• functionality to ignore multiple dirty propagations from same source by @k-dominik in #2694
  • this enables faster stage 1 retraining in autocontext if stage 2 has already been trained
• release pin on pytest by @k-dominik in #2633
• Add Trainable Domain Adaptation workflow by @k-dominik in #2704
• Rearrange labeling UI by @btbest in #2705
• fix tda export names by @k-dominik in #2708

Full Changelog: 1.4.0...1.4.1b1

1.4.0

Highlights

• Neural Network Workflow: run pre-trained neural networks from Bioimage.io Model Zoo in ilastik
  • Networks can run on CPU, GPU (choose the GPU build for Windows, or Linux), or on a remote server
  • No Neural Network training, yet - we’re working on it.
• New startup screen!
  • Major UI and performance improvements for the Multicut Workflow
  • Stability/Usability improvements in the Carving Workflow
  • By popular demand: redo/undo for brush-strokes
  • Many bugfixes throughout

Downloads

Please visit our website for more information.

• Windows
• OSX
• Linux

GPU-enabled builds (will make running the Neural Network Workflow faster if you have a compatible nvidia GPU in your machine):

• Windows
• Linux

ilastik can also be installed via conda/mamba, see our Readme.