Semantic segmentation in an urban area can be utilized to differentiate between various objects on LiDAR point cloud data. This research aims to distinguish between buildings object and non-buildings object by performing semantic segmentation on the LiDAR point cloud data. A deep learning method has been proven to achieve state-of-art performance on semantic segmentation task. Dynamic Graph Convolutional Neural Network (DGCNN) is used to perform semantic segmentation in this research. Two datasets from two different regions are used to perform semantic segmentation. The first dataset is retrieved from Margonda region in Depok, Indonesia, and the second dataset is retrieved from Dublin region in Ireland. The experiment shows that the deep learning method is capable of doing semantic segmentation on LiDAR point cloud data. When tested the first dataset achieved accuracy of 86,3% and mean IoU of 70,3%. The second dataset achieved accuracy of 81,9% and mean IoU of 65,2%.