Hi,
I’m using the interactive labeling editor frequently and realised that it would really make things much easier if there would be a faster way to select labels in the “current label” list. A very common use-case is the merging of two (or more) over-segmented labels.
So far you first have to look up the label ID of one of the two labels you want to merge in the image then go to the “Current label” list on the right to find and select this label (which takes quite some time in case you have many labels). Then you can go back to the image delete the label you don’t want to keep by right mouse click and adding the selected label from the list by left mouse click.
It would speed up the whole process a lot if one could instead of manually scrolling through the list to find and select your label of interest do the same directly in the image. Left or right mouse click wouldn’t work here of course because that assign or removes labels but I was thinking of a hotkey that would automatically select the label from the “Current label” list that the cursor is currently pointing at in the image. Maybe there is already such an option and I’m just not aware of it but if not it’s probably not too difficult to add it.
Hi @COS_HD1 / Christian,
it looks like your are struggling with this node called Interactive Labeling Editor from the Image Processing Extension.
Did you check out the Labeling View new node? It came out with 4.1 in December. It works via an interactive JavaScript view to let you label images. It is part of the active learning extension but you can use it to label images also in a different scenario.
Thanks for your suggestion. I was not aware of the new labeling view node. However after I tried to use it I’m not quite sure if this would work to edit multidimensional image labels. What I would like to do is to edit a three dimensional image labeling (which is basically a segmented image). As far as I can see from the example workflow you suggested one can just assign labels or classes to (2D?) images with the labeling view node. If I would like to merge two segmentation labels I don’t see how this is going to work with the “labeling view” since it doesn’t even accept image labels (datatype: SEG) as input. I think the labelling editor is the right tool for the task and I think it is quite nice. I just think it could be improved a lot by adding the function that I suggested in my first post.
Thank you for your suggestion, it seems indeed like something we should evaluate adding to the interactive labeling editor. I can understand how it would improve your productivity with the node.
I can’t promise a date when this would be available though.
