Roi manager vs. labeled image (imagej2 way to handle roi's)


it’s been some time I inquired about the imagej2 way of handling regions of interest.

now I wanted ask again about the status of things:
Let’s say I got an image, segement it with any given algorithm to generate a binary mask and then want to identiry the different regions of interest (label it). With the labelled image I then want to compute measurements for the individual roi’s. Now this is well documented with the ROI manager. The problem here is that the ROI Manager and a ResultTable, that both are attached to their UI. Furthermore this extends badly to volumetrics.
What I am looking for is something to

There is the 3D suite, which does nice work using the UI
… but it seems to me this is not necessairly a good basis to develop new plugins on top of.

My principal question is how to implement a plugin using without the ROI Manager and using new data structures like Dataset that could be run headless?
Is this even possible yet? and if so, what is a good starting point?

1 Like

Possible, yes. But the API is not finalized. In January 2015 the ROI API of ImgLib2 was redesigned. My understanding is that that design is not yet finished—@tpietzsch knows the details.

Regarding labelings specifically, @dietzc is the master there. Christian: is there any sample code you can point at for how things are currently being done with labelings?