search

knime-ip / knip

KNIME Image Processing Extension
https://www.knime.com/community/image-processing
49 stars 11 forks source link

Image Viewer: Label values of 0 are indistinguishable from empty labels #463

Open imagejan opened 7 years ago

imagejan commented 7 years ago

In the Image Viewer node, labelings that have a label value of [0] (that you get e.g. when using a Create Labeling aggregation and using an Integer column for the labels) are shown in white and are indistinguishable from the EmptyLabel white background. Also, overlapping labels [0; 1] are indistinguishable from label [1].

The Randomize Label Colors function doesn't change the white color of 0 labels.

In the Interactive Segmentation View, everything works as expected (even without using an Image column), i.e. 0 labels have random colors.

tibuch commented 7 years ago

Unfortunately I can't reproduce this bug.

What I did:

  1. Add a label with Interactive Annotator.
  2. Use Label Transformer and set label to 0. (probably still a string label)
  3. Crop label with Segment Cropper.
  4. Use GroupBy and create a labeling with inttype labels.

Result in Image Viewer: image

imagejan commented 7 years ago

Ok, I tried to build an example workflow and encountered yet another related bug.

Here's my workflow: Label 0 Bug.zip

When I use Annotator > Feature Calculator > GroupBy with Integer labels, the labeling with value 0 extends to the full image dimensions:

image

When joining labels with offsets from other sources, I get the behavior described in the above post:

image

tibuch commented 7 years ago

Wow! That is "fascinating". I will investigate. Thanks for reporting.

imagejan commented 6 years ago

Any news on this? The issue is still the same with KNIME Image Processing 1.6.1.201806190602.

imagejan commented 2 years ago

@tibuch did you ever get around to investigating this? 😉 😆