Tree Point Classification not editing classification
Hi everyone,
The short version is that when I run 'Classify Point Cloud Using Trained Model' on a .lasd file using the tree point classification deep learning pack, the point cloud is unaltered; the point classifications remain the same. The output indicates a success and there are no errors or warnings.
Here's the long version:
I'm on ArcGIS Pro 3.2.2, Windows 11 with an NVIDIA GeForce RTX 3060 graphics card. I've obtained a bunch of compressed LiDAR files (.laz) which I've converted to .lasd using the 'Convert LAS' tool. I keep all the default settings for this tool. I then load the .lasd file and see what I expect to see as per the metadata from the data provider. The data comes with the following classifications: 1 Unassigned, 2 Ground, 7 Noise, 9 Water, 10 Rail, and 18 High Noise.
As a brief aside, I can use tools such as 'Classify LAS By Height' and 'Change LAS Class Codes' with success. For instance, 'Classify LAS By Height' worked quite effectively to identify low, medium, and high vegetation using the ground points as the reference surface. Unfortunately buildings are consistently bundled with high vegetation, hence the need to use the tree classification tool.
I downloaded the tree classification pack and used it in the 'Classify Point Cloud Using Trained Model' tool numerous times, changing minor details here and there. A few of the things I've tried: I made a LAS dataset and loaded of the .las files into this dataset and used it in the tool, but I've also used the individual .las files as well. I've tried turning on each one of the target classification classes, using a processing boundary, downloading the tree classification package once again. I updated ArcGIS Pro to the latest version, redownloaded the deep learning installer for Pro 3.2, and restarted my computer. Every time I run it I get the same success message from the tool (screenshot below).
I'm very confused and unsure what I'm missing. Does anyone have any suggestions?
