Best way to display additional information as overlay?

Hello everyone,

I am currently trying to display additional information as an overlay on top of an image with the goal to provide a human annotator with additional visual information.

Since the information I want to display is currently only provided on a relatively coarse rectangular grid, I am creating individual Detection Objects with a measurements attached. In the future this might be measurements on a per pixel basis.

Is there a better way to display an image overlay that the user can easily toggle invisible?
I feel as if ideally I would want to display another RGB image on top of the current image for which the user can select colormaps and adjust transparency. But because currently my calculated data is on a relatively coarse grid the DetectionObjects seem to do what I need.


Assuming 0.2.0+ since I can see the Classifier button:
If you have a second image, have you tried the Interactive image alignment in the Analyze menu? If your images are already aligned, that simplifies things, and you could just use the slider to alternate which image was visible. It isn’t as clean as a button click, but you can also shade things 50/50 to see both at once (as you did with the measurement maps).

Though with the detections you do have the D key, which is a bit faster.

Maybe you want something similar to what was done here:

1 Like

@poehlmann the detection objects might be the best way to go, since the detections can be stored alongside specific images.

An alternative is to create a BufferedImageOverlay:

It was written for the kind of purpose you describe, however making it practical could require a lot more code (e.g. adding listeners to handle image changes, reading/writing grids if required). The BufferedImageOverlay therefore mostly exists to be used by future commands… but it’s there if you need it for advanced scripts as well.

1 Like

Thank you @Research_Associate and @petebankhead !

I’ll have a look at the Image Alignment and the BufferedImageOverlay. Both seem to be able to do what I will need further down the line :smiley:

Currently the DetectionObject solution seems to work fine. I have so far tested this with about ~100000 DetectionObjects and haven’t tried how far I can push it.

I might have to see if I can do something about the selection issue when DetectionObjects are smaller compared to Annotations (Double click selects detections even if hidden?) and if there’s a way to display them filled but without border.

Thanks again for your help!

1 Like

Potentially yes if you set the detection line thickness to 0 in the preferences; but you might not need to if you create tile objects rather than detection objects (where tile is just a special case of detection).

1 Like

Awesome! That’s exactly what I needed :smiley:

1 Like