top of page

Point cloud segmentation with deep learning

In this project, I will demonstrate how to train a Point Net on an example building model for semantic segmentation. PointNet is a new architecture that can do classification and segmentation tasks by utilizing full point clouds. Unlike previous approaches for organizing data, PointNet accepts points exactly as they are, with no requirement for grids or images. This makes it extremely essential in the world of 3D.



Result






bottom of page