 This paper compares three different methods for ground classification from 3D point clouds, a heuristic method, a deep convolutional network based on SegNet that classifies 2D images generated from the 3D point cloud, and a deep learning classification based on point net that takes 3D points directly as inputs. The results show that deep learning based approaches outperform the heuristic method with F scores above 96% and the best results were obtained using a shallow version of SegNet with an F score above 97%. This article was authored by Mario Soelen, Belen Raveiro, Hesius Balladou and others.