Pixel Classification to Object Classification - not compatible format


I’m still trying to get the hang of using Ilastik, and I was hoping I can get some help in trying to now use my pixel classification data to the object classification features. My goal is to count nuclei in 3D across an image file that is about ~100 z stacks. I train it through the pixel classification and I aim to get segmentation data that is exported. When I go to use the object classification, I add my raw data file (which is already in the Ilastik hf format that I changed through Fiji), and when I want to add my segmentation file, I get an error code saying that my data is not the correct format or that the file I chose had multiple partts to it (lots and lots of labels etc).

I was hoping to get some help on what I’m doing wrong/ ways I can make these two workflows work for me.

Hello @Yoko_T,

we recommend saving the Probabilities image from ilastik pixel classification, for use in the Object classification workflow (Raw Data, Pixel Prediction Map).
You can do some nice things in thresholding there to get the final object segmentation and then train a classifier on top of these objects. (See also our docs for object classification).