Today after Esri UC2025 conference I get a chance to explore updated (July 9,2025) Tree point classification deep learning package with my "lasd" data set. While the model successfully identified trees, it showed limitations in accurately detecting tree canopies even within the urban context it was trained on. In several instances, trees and buildings were grouped together, resulting in less precise segmentation and affecting the reliability of the output with buildings appeared as trees in some places. Any suggestions for the improvements.
