Despite multiple attempts, the cgal point classification code fails to run on anything other than the demo dataset. I have attempted with multiple .ply files with a scalar field containing training labels.
The point datasets are around 2 million (which is relatively modest), I have attached a link to one of them https://drive.google.com/file/d/1DRvn75_3F_AQNrkSf2fBDUeZr8wld7Qk/view?usp=sharing.
The seg-fault occurs using both C++ & python. Given the video demo on the cgal website shows a larger cloud processed this seems strange.
Any cgal developers know why this may be?
Thanks to cgal team this has been resolved. The issue was with the numbering of my labels scalar field (for the the training of the algorithm) of the ply file which was not contiguous.
To clarify, the field properties of the ply file should be as follows for use with the classification functions...
x double y double z double red uchar green uchar blue uchar nx double ny double nz double label int
...with the last field "label" contents (i.e. the class labels) be something like -1 0 1 2.