Segmentation fault during creation of second pass about openpointclass HOT 5 CLOSED

Sure, here you go: https://www.dropbox.com/scl/fi/x80gaacczqmfyhlj8574j/21A027_628_58_0000.zip?rlkey=kxi6noiaybo7hyd5h4qik1xri&dl=0

If you prefer another delivery method just let me know, I can't upload it as an attachment since the file limit is 25MB. I got the data from the Swedish government, as it is publicly available. The files they are serving are about 50MBs and I suppose it is the unmodified (no lastools etc) version that you want.

Thanks in advance.

from openpointclass.

Seems to be a crash due to the lack of colors in the input.

I've just added 0f23e58 which sets a default color if there are no colors in the input point set. Note that you will likely need to train a new model on non-colored data for it to work.

Try to rebuild/retry? 🙏

from openpointclass.

Can you share the .laz file? I could try to take a look.

from openpointclass.

Works like a charm! Thank you very, very much.

A final note on my end is if you have any recommendation with regards to "training a new model on non-colored data"?
What I have found so-far are the examples provided in this repository: https://github.com/OpenDroneMap/ODMSemantic3D
If the approach in the ODMSemantic3D is the recommended approach, then, how much data would be sufficient and how accurate would I need to be.

Again, thanks a bunch!

from openpointclass.

The consensus with these classifiers is that the more data, the better, although at some point you'll have diminishing returns. Glad it worked!

from openpointclass.