This project presents a system that captures and stores an accurate 3-dimensional map of an arbitrary environment. Building on KinFu, an open source implementation of KinectFusion, a GPU- accelerated volumetric surface reconstruction algorithm, within PCL and SLAM++, an algorithm for localization and camera tracking, this project seeks to automate the recognition process. Model matches and orientations are computed via an algorithm that matches points and surface normals of the scene with those of a database model. This process is ideally suited to a parallel implementation, potentially allowing real-time object recognition.