Run HAIS with own data

[LaureenK]

Hi, I would like to run HAIS on my own dataset. I saw that you got fixed values for the ScanNet Dataset inside your implementation: For example Hierarchical_aggregation.cpp:

// instance point num for each class, statistical data from the training set
float class_numpoint_mean_dict[20] = {-1., -1., 3917., 12056., 2303., \
                                    8331., 3948., 3166., 5629., 11719., \
                                    1003., 3317., 4912., 10221., 3889., \
                                    4136., 2120., 945., 3967., 2589.};

Which files do I have to change (except the data loader) to get it run with my own data? Thanks for your help!

[outsidercsy]

You need to modify the definition of class_numpoint_mean_dict (Line 7 in hierarchical_aggregation.cpp) and class_radius_mean (Line 55 in hierarchical_aggregation.cu), according to the statistics of your data.

[SijanNeupane49]

How can one calculate class_numpoint_mean_dict and class_radius_mean for own data?

[meidachen]

How can one calculate class_numpoint_mean_dict and class_radius_mean for own data?

I was using the following script for my own data. You may need to modify it to make it work for yours. https://github.com/meidachen/STPLS3D/blob/main/HAIS/data/prepare_data_statistic_stpls3d.py