Point Cloud Classification

[LITA_232]

Hello everyone,

I have an indoor point cloud from Leica BLK360. It is unclassified. I'd like to classify it. What would be the best way to do it?

[infrared_sight]

Hello, it depends, what type of features would you like to classify?

[smacl]

Some fantastic tutorials on automated point cloud segmentation for building internals here https://towardsdatascience.com/how-to-a ... 3c9039e4f5 You'll need to dip into some Python but could be an interesting journey.

[Kruse]

Hello everyone,

I have an indoor point cloud from Leica BLK360. It is unclassified. I'd like to classify it. What would be the best way to do it?

If you're running the latest Register 360+ from Leica, it has some autoclassification tools built into the software now (after downloading the classification install file and assuming you have a new enought GPU to run the application). While it's not the fastest option, it works decently. You can also manually classify areas of the point cloud if it didn't auto classify correctly.

[LITA_232]

Hello, it depends, what type of features would you like to classify?

I'd like to try to extract as many indoor assets as it possible: light switches, fire alarms, lamps etc.

[LITA_232]

Some fantastic tutorials on automated point cloud segmentation for building internals here https://towardsdatascience.com/how-to-a ... 3c9039e4f5 You'll need to dip into some Python but could be an interesting journey.

Thank you for the answer. Yes, I was looking at this tutorial as well. But I noticed that, for example, sometimes the algorithm classified door handles as one class (one color), but not all (different color). Ideally I'd like to have a point cloud where all (as many as possible) indoor assets are combined in classes