CloudCompare forum

Hello, I've been trying to do some classification tests with 3DMASC in cloud compare to see if it does a better job of identifying Coarse Woody Debris (deadfall), Vegetation, and bare earth in a forestry scene than Canupo. Unfortunately, even after after following the youtube tutorial (https://www.youtube.com/watch?v=6z40FLN ... HV&index=8) and reading the documentation (https://lidar.univ-rennes.fr/node/11) I still seem to be struggling with the config file.

I am able to open the Train Classifier dialogue to choose the PC1 cloud, but after clicking OK I receive a popup, and a message in the CC console, which states "Failed to load the training file"

Here are the parameters I am working with, any help figuring out this issue would be most appreciated!



# NAMES
cloud: PC1= #Point Cloud for training, leave as = so you can choose a cloud loaded in Cloud Compare

# CORE POINTS
core_points: PC1

# SCALES
scales: 0.1;0.5;1;1.5 #scales as a list

# FEATURES
feature: INT_SC0_PC1
feature: Z_SC0.5_PC1
feature: ZMAX_SCx_PC1
feature: ZMIN_SCx_PC1
feature: ZRANGE_SCx_PC1
feature: SPHER_SCx_PC1
feature: LINEA_SCx_PC1
feature: PLANA_SCx_PC1
feature: ROUGH_SCx_PC1
feature: CURV_SCx_PC1
feature: ANISO_SCx_PC1

As a quick follow up on the main post, for if/when this gets reviewed and made public, I was doing some more testing this morning and I figured something out, namely that the issue I was having related to these lines:

feature: Z_SC0.5_PC1
feature: ZMAX_SCx_PC1
feature: ZMIN_SCx_PC1
feature: ZRANGE_SCx_PC1


I found this out by going through the console logs from when I was attempting to load the point cloud. Specifically, this was the warning/error message in the console: "[10:40:03] Malformed feature: scaled point features need a STAT measure to be defined (line 12)".

Line twelve is this code: "feature: Z_SC0.5_PC1", so I decided to simply remove that line and try again, only to receive the same message. However, when removing all of the above lines relating to the Z I was able to get the parameter file to load properly with my point cloud.

So, while my issue has been "solved", for a given definition of the word, I would still like to include those parameters in my classification. Therefore, if anyone has any advice as to how I can go about computing the necessary measurements to get them working I would be most grateful.

The message is (almost) explicit, "scaled point features need a STAT". Only one line is wrong in your parameter file :

feature: Z_SC0.5_PC1

should read (replace STAT by any tag among MEAN MODE MEDIAN STD RANGE SKEW)

feature: Z_SC0.5_STAT_PC1

as specified in

https://lidar.univ-rennes.fr/sites/lida ... syntax.pdf