Feature detection on point clouds

[solarjoe]

Hello,

I have a question and hope that you can help me.
I need to apply 3D feature detection techniques to point clouds to process Lidar
data for automated driving. Since I am familiar with CC I would like to
preprocess the data there.

The idea is to have an approach like YOLO on 3D data and get results like
the ones shown in https://miro.medium.com/max/1400/1*doz4 ... 29FIHw.png

I read about the CANUPO addon, but I am quite sure that this won't work for my
task. There seem to be several algorithms available for this kind of task
(PointNet, PointNet++, LaserNet, Point Pillars, Complex Yolo).
Has one of you worked on a similar task with CC before and has an idea
how a workflow with these algorithms could look like? Are there maybe addons available?

Thanks,
Joe

[daniel]

I personally don't have any experience with these algorithms, and I guess you would need to create your own plugin to use them in CC?

(unless someone already did it of course ;)