This document discusses using deep learning techniques for semantic segmentation of indoor point clouds. It provides an overview of initial ideas for using deep learning models trained on 3D CAD models to classify and label points in an indoor point cloud. It also discusses pre-processing the point cloud through techniques like denoising, upsampling, and finding planar surfaces to simplify the input before semantic segmentation. The order of semantic segmentation and 3D reconstruction is noted as something that could potentially be swapped.