Skip to content

Release Notes

Jump to bottom Edit New page
Dominik Kutra edited this page Oct 12, 2020 · 29 revisions

ilastik 1.4.0b8 (12. October 2020) non-public release

  • note: it is currently recommended to delete your old .ilastik_preferences file located in
    • win: C:\Users\<USERNAME>\
    • osx: /Users/<USERNAME>/
    • linux: /home/<USERNAME>/
  • preferences now saved as json
  • changed default path for logs:
    • win: C:\Users\<USERNAME>\AppData\Local\ilastik\Logs
    • osx: /Users/<USERNAME>/Library/Logs/ilastik
    • linux: /home/<USERNAME>/.cache/ilastik
  • changed default path for preferences file:
    • win: C:\Users\<USERNAME>\AppData\Local\ilastik
    • osx: /Users/<USERNAME>/Library/Caches/ilastik
    • linux: /home/<USERNAME>/.config/ilastik
  • Updated Neural Network Workflow
  • Added viewer caches
  • Added Undo/Redo for brush strokes (ctrl+z, ctrl+y)
  • fixed range adjust for layers
  • fixes to data selection
  • updated dependencies:
    • Qt 5.6 -> 5.12
    • boost 1.70 -> 1.72
    • sourcing vigra and lemon from conda-forge

ilastik 1.4.0b1 (2. March 2020)

  • added the Neural Network Classification workflow 🚀 🚀 🚀. Se also the tiktorch repository
  • improved multicut preprocessing
  • dependency upgrades

ilastik 1.3.3post2 (13. December 2019)

  • fixed compatibility issue in the PC+OC workflow
  • Object Classification: make available object identities for export.

Downloads

ilastik 1.3.3post1 (26. October 2019)

  • fixed some compatibility issues
  • better default directory (placeholders) for saving of object table
  • fixed data selection drag-n-drop

Downloads

ilastik 1.3.3 (7. October 2019)

Downloads

ilastik 1.3.3rc2 (30. September 2019)

  • rework of data selection
  • counting fixes (saving boxes, import/export of boxes, interaction improvements)
  • drag-n-drop in batch processing
  • more efficient memory utilization in carving
  • a lot of internal backend cleanups/improvements
  • a lot of internal GUI cleanups/improvements

ilastik 1.3.3b2 (6. June 2019)

New features:

  • applet bar improvement, redesign
  • carving improvements

Downloads:

ilastik 1.3.3b1 (13. April 2019)

New features:

  • Support for n5 volumes (https://github.com/saalfeldlab/n5)
  • Support for masks in Pixel Classification (an additional dataset can be supplied that disables computation of probabilities wherever it is 0 -> speeds up computation in cases with a lot of background
  • Atlas Mask overlays for Object classification (an additional dataset can be supplied. Pixel values of this additional mask are included in result tables but not used in classification.)
  • autocontext now takes arbitrary data types (was previously limited to uint8), but mind the memory footprint! We still recommend to use uint8.
  • Lots of internal fixes and an updated dependency stack

Downloads:

ilastik 1.3.2post1 (7. March 2019)

  • fixed a regression in tracking export

ilastik 1.3.2 (7. February 2019)

Summary of addition since 1.3.0

  • New color way!
  • Image features: sigmas can be specified, 2D/3D filter selection
  • Training data no longer required for headless processing
  • Carving improvements: block-wise preprocessing, various improvements, fixes
  • Window-leveling available in most workflows
  • Various improvements and fixes to the object classification workflow
  • Two labels always added in Pixel and Object Classification
  • Fixed Fiji-MaMuT export plugin
  • Will even work on Halloween ;-)

ilastik 1.3.2b3 (20. September 2018)

  • New color way!
  • Overall simplification of the image feature computation
  • Custom selectable sigmas for image features
  • (Given 3D data) Image features can be selectively computed in 2D or 3D
  • Training data no longer required for headless processing
  • Various improvements and fixes to the object classification workflow
  • fixed counting index error

ilastik 1.3.1b1 (22. March 2018)

  • Carving improvements
    • block-wise preprocessing
  • various fixes
  • Linux: dependencies now built with gcc5;

Note to ilastik developers on linux: We strongly advise to create a new conda environment now

ilastik 1.3.0 (26. January 2018)

  • First stable ilastik release based on python 3 and QT 5
  • supply stacking axis in headless mode by supplying the --stack_along command line option

ilastik 1.3.0b4 - BETA (22. December 2017)

  • fixed loading and saving of data and projects for files with non-Latin characters
  • for the Multicut workflow on Windows, fixed the lookup of commercial solvers
  • updated dependencies (vigra, yapsy, nifty, multi-hypotheses-tracking)

ilastik 1.3.0b3 - BETA (24. November 2017)

  • fixed carving import
  • fixed some byte <-> string <-> unicode conversion problems which affected loading of previously saved Multicut workflows.

ilastik 1.2.2post2 (24. November 2017)

  • fixed carving import
  • forward compatibility - import projects saved in 1.3.x

ilastik 1.3.0b2 - BETA (20. November 2017) - Linux only

  • fix problem in the linux binary that made keyboard input impossible due to some missing QT_XKB configuration (fixed by setting it in run_ilastik.sh and rebuilding the ilastik-launch package)

ilastik 1.3.0b1 - BETA (17. November 2017)

  • in the transition from 1.2.2 to 1.3 we have updated all our dependencies, switched from Python 2 to 3, updated from Qt4 to Qt5 etc
  • the switch from Python 2 to 3 can break compatibility with old project files due to a change in string/byte handling
  • dependencies are now built as conda packages using the same code and setup on all 3 operating systems
  • Tracking finally uses the same backend on Windows so that it is available without solvers on all platforms
  • Tracking now provides many export formats on all platforms, e.g. CSV, Fiji's MaMuT, Cell Tracking Challenge formt, ...
  • Edge-based segmentation workflow available without solvers on all platforms
  • Edge-based segmentation and tracking workflows obtain additional functionality if CPLEX 12.7.1 or Gurobi 7.5.1 are installed
  • 3D display of the slicing planes and the meshes in the Carving workflow now based on pyqtgraph instead of VTK (removes a big dependency)
  • 3D convex hull features available for objects in the Object classification and tracking workflows
  • Uncertainty layer for Object classification that helps to locate and annotate objects where the classifier needs guidance.
  • Many more small fixes and enhancements
    • fixed WYSIWYG export (the camera icon)
    • integration testing on Windows (AppVeyor) and linux (CircleCI)
    • more tests
    • many Python3 and Qt5 related fixes
    • slot declarations are cleaned up
    • lots of legacy code removed

Known issues:

Add a custom footer
Add a custom sidebar

Clone this wiki locally