Core points

[Madeleine]

Hi,

Thank you for this powerful tool that helps me with my master's project. Nevertheless, I still have some questions:

I noticed even though I created my classifier using the command line: PC1_SS_R0.5 to use only half of the points in the cloud as core_points, this command line disappears when I save the classifier (after the training period). In other words, when I want to use my classifier during the classification procedure, 3DMASC uses all the points in the cloud I want to classify, instead of just half the points (core points). I manually changed the line about core points to include the random subsampling part (_SS_R0.5), but 3DMASC still uses all the points as core points. I don't understand why.

Is that normal?
Is it possible to use core points even with the classify procedure, or is it just for the training procedure?

Thank you very much for the help,

Madeleine

Here is my Classifier file (bold = not written when I exported and saved the classifier, but added manually):

# 3DMASC classifier file
classifier: Classifier5.yaml
# Clouds (roles)
cloud: PC1
# Core points (classified role)
core_points: PC1_SS_R0.5
# Features
feature: SPHER_SC0.4_PC1
feature: SPHER_SC0.6_PC1
feature: SPHER_SC0.8_PC1
feature: PLANA_SC0.4_PC1
feature: PLANA_SC0.6_PC1
feature: PLANA_SC0.8_PC1
feature: PLANA_SC1_PC1
feature: ROUGH_SC0.6_PC1
feature: ROUGH_SC0.8_PC1
feature: PCA2_SC0.4_PC1

[paul.leroy]

The classifier do its job: you give it points, it classifies them.
So you have to subsample your cloud if you want to classify only a fraction of it.