ArcGIS Pro

Classify a point cloud with deep learning in ArcGIS Pro

Did you know that deep learning can now be applied to point clouds to aid in classification? Here are some links and pointers to resources that will help you learn about this new technology.

There are deep learning classification models available at ArcGIS Online, such as the Tree Point Classification model and Power Line Classification model. These models were trained by expert data scientists. If these models are appropriate for your project, you can use them to classify your point clouds. If these models are not suitable for your project, you can train your own models from scratch using your own training data, or use your data to tune the pretrained models.

A Learn ArcGIS lesson, Classify power lines using deep learning, has been created to help familiarize you with the deep learning workflow of point cloud classification in ArcGIS Pro. It uses power line classification as an example to teach you how to train a deep learning classification model from scratch and how to use the trained model to classify power lines from point clouds. The lesson gives you instructions on installing deep learning libraries, checking dedicated GPU memory usage, setting parameter values, and evaluating the quality of the trained models. Even if you’re not interested specifically in power line classification, the concepts taught in the lesson will help you learn how to use the technology for other applications.

 

Requirements

Related topics

Introduction to deep learning and point clouds

Train a deep learning model for point cloud classification

Classify a point cloud with deep learning

Access point cloud training results

 

Banner picture created from data provided by NASA Grant NNX13AP69G, the University of Maryland, and the Sonoma County Vegetation Mapping and LiDAR Program.

About the author

Jie is an Esri Software Product Engineer for 3D Analysis. He has over 10 years of experience in 3D, imagery, spatial analysis, enterprise geodatabase management, and Web GIS. Jie received his Ph.D. in Geospatial Information Sciences from the University of Texas at Dallas in 2011.

0 Comments
Inline Feedbacks
View all comments

Next Article

Changing Directions

Read this article