GeoCue’s Jaume Cazorla shows how AI classification in LP360 can help turn large point clouds into usable data for mapping, forestry, and utility projects.
At INTERGEO 2026, GeoCue Customer Success Manager Jaume Cazorla presented “What Can AI Do in LiDAR Processing?” on behalf of his colleague Thadeu Lugarini, GeoCue’s Customer Success Manager for South America, who organized the session. Jaume walked through how AI classifies LiDAR point clouds in LP360, from isolating bare earth for contour maps to identifying individual trees and extracting power lines.
Watch the full presentation below, or keep reading for the highlights and examples from the session.
Watch “What Can AI Do in LiDAR Processing?” on YouTube:
From classifying points to checking results
With better scanners capturing more detail across larger areas, processing can become the bottleneck. In a traditional classification workflow, users select tools, set parameters, run algorithms, and then review and refine the results. With AI classification, a trained model handles the initial pass, giving users more time to check the results and correct areas that need attention.
Jaume explained how the team builds these models using manually classified point clouds. The team trains a model on multiple datasets, tests it on raw data, and uses the errors it finds to improve the next round of training. The data source matters as much as the task. A model trained to identify trees in airborne LiDAR, for example, may see very different point patterns in a scan collected from the ground.
What AI classification looks like in LP360
LP360 now offers four new AI Classification tools designed for different LiDAR workflows. In his INTERGEO presentation, Jaume showed how each tool works with real point cloud data, from mapping bare earth to identifying individual trees and power lines.
AI Ground identifies bare-earth points beneath vegetation so users can build digital terrain models and generate contours. Jaume showed a wooded site where the ground was difficult to distinguish in the original point cloud.
AI Ground+ expands classification beyond bare earth. In the urban examples, Jaume showed how the workflow separates vegetation and features such as roofs, walls, fences, and vehicles into classes users can review independently.
AI Forestry separates ground and vegetation and identifies individual trees. Jaume also showed how LiDAR can support diameter at breast height (DBH) measurements when the scan captures enough points on the trunk.
AI Utilities identifies power lines, supporting structures, and nearby vegetation. Those classes give teams a starting point for checking how close vegetation is to the lines and identifying potential encroachment.
Where AI classification goes next
The examples in the talk focused mainly on airborne LiDAR. Jaume also discussed extending the models to data from mobile mapping systems, handheld scanners, and terrestrial laser scanners. Each of those capture methods sees a scene from a different perspective, which affects what a model can reliably identify.
Jaume also shared where AI fits into his own workflow. Once he’s satisfied with the point cloud geometry, he sends the data for cloud classification and reviews the results when they come back. Upload and download times factor into turnaround, especially for large datasets. The time savings come from automating the initial classification so he can spend more of his time checking the output and refining the deliverables.
LP360 continues to explore new ways to modernize LiDAR processing with AI classification and other AI-powered tools, so you can spend less time classifying points and more time reviewing the results. Ready to try it on your own data? All four AI Classification tools are available now in the store at LP360.com. Choose the tool that fits your next project and add it to your workflow.