What reasons can cause a failure in AI Automatic Classification?

The most common failures are usually due to lack of performance in the instrument with respect to the Point Cloud to be classified. Four tips for successfully classifying the point cloud are:

• Always perform a cleanup of the cloud (removal of external noise points, Zones that are not of interest, …) quickly using the selection tools.

• Perform a Density Selection at 0.01m, hiding or eliminating those selected points, thereby reducing RAM memory usage during the process.

• Partition the point cloud using the selection tools (Polygon Selection, for Example) to obtain smaller sections of clouds using the Filter tool, aiming to run the classification on smaller sets.

• If the error occurs during the "Classifying" stage of the process, try to reduce the Radius parameter of the AI Automatic Classification. This Value is determined based on your instrument, but if the Point Cloud proves to be very volumetrically dense, such as resulting from a scan over a wooded Zone, the indicated Value may exceed the capabilities of your instrument and significantly slow down the classification, to the point of causing a failure.

Additionally, if you receive the error while performing a classification with the Exterior Model (10cm), you can try using the Exterior (25cm) Model, which will be less precise at the intersections of Details with different categories, but will require less computational power and significantly shorter times.