How to use the label_class_selection.prototxt

[TITC]

How to use this prototxt file, I found the fist time it appeared is caffe_nd_sense_segmentation

and I also noticed that the author has participated in cdeep3m cdeep3m contributor

I try to adjust the label and data layer for some improvement, but I can't find any descriptive information about the model input layer PachData layer and the usage of this prototxt file?

label_class_selection.prototxt

  ignore_rest_of_label:true
  rest_of_label_mapping:false;

  rest_of_label_prob : 0.5;

   label_prob_mapping_info{
         label: 2
         prob: 0.3
 }


  label_prob_mapping_info{
         label: 15
         prob: 0.11
 }