How to implement multilevel classification in eCognition (segmenting imagery at different scales)

I am trying to perform a ground cover classification in eCognition using UAV RGB imagery of a forest ecosystem, with multiresolution segmentation algorithm and NN classification.

Since there are terrain objects that are better segmented on a scale parameter = 200, and others on a scale parameter = 70, how could I implement a classification that would integrate objects segmented at both levels (i.e. Level 1 = 200 and Level 2 = 70)?

This is exactly what they do in these papers:

• https://doi.org/10.1016/j.rse.2011.11.020
• https://doi.org/10.1016/j.rse.2010.12.017
• https://doi.org/10.1016/j.jag.2011.06.008

I am classifying at one level, and then applying a mask to classify at another level so perhaps ignoring the masked areas could work.

I am not able to implement masks in classified objects.