Skip to content

Releases: fiji/Trainable_Segmentation

Trainable_Segmentation-3.2.26

21 May 09:22
Compare
Choose a tag to compare

Major changes:

  • Fix bug in the hyperstack formed when creating a probability map from macro (reported by Arturo Ibáñez).

Trainable_Segmentation-3.2.25

18 May 14:01
Compare
Choose a tag to compare

Major changes:

  • Add method to apply a classifier on an image using tiles.

Trainable_Segmentation-3.2.24

13 Apr 15:04
Compare
Choose a tag to compare

Major changes:

  • Fix bug that allowed using GUI while saving a classifier to file.

Trainable_Segmentation-3.2.23

28 Mar 09:16
Compare
Choose a tag to compare

Major changes:

  • Fix bug on the calculation of ImageScience features with anisotropic voxel size.

Trainable_Segmentation-3.2.22

06 Mar 09:23
Compare
Choose a tag to compare

Major changes:

  • Fix sigmas in 3D features so they are in voxel units and are adjusted based on the image calibration.
  • Fix bug in 3D structure feature to store middle and smallest eigenvalues correctly.
    Special thanks to James Lefevre who pointed out the bugs!

Trainable_Segmentation-3.2.21

26 Jan 11:09
Compare
Choose a tag to compare

Major changes:

  • Fix bug when using class names without samples.

Trainable_Segmentation-3.2.20

20 Nov 10:57
Compare
Choose a tag to compare

Major changes:

  • Fix bug that was preventing the settings dialog to be displayed after loading a 3D classifier.

Trainable_Segmentation-3.2.19

14 Nov 10:43
Compare
Choose a tag to compare

Major changes:

  • Fix bug when loading a classifier trained on features that did not use a sigma.

Trainable_Segmentation-3.2.18

10 Nov 09:47
Compare
Choose a tag to compare

Major changes:

  • Allow adding 3D samples programmatically from binary labels.
  • Make sure label output is 8-bit.
  • Replace autoThreshold() calls by correct thresholding.

Trainable_Segmentation-3.2.17

20 Sep 11:00
Compare
Choose a tag to compare

Major changes:

  • Fix bug when using a 3D classifier from script without the GUI.
  • Set correctly probability map slice names in all methods.