Media

PCL became a first-class citizen project in March 2011, when Radu B. Rusu, while working as a Research Scientist at Willow Garage, decided to create a separate standalone initiative for 3D point cloud processing and host it under pointclouds.org. Some of the first algorithms that form the foundation of PCL today have been developed independently by many research groups around the world, and have been slowly transitioned from publications into code by the PCL community. For example, three of our current PCL developers (Nico, Radu and Zoltan) have worked together under the supervision of Prof. Michael Beetz at TUM in Germany on multiple 3D topics between 2005-2010, which led to a good understanding on how a project like PCL should be structured later on.

The video shown on the right side of the screen presents the history of the svn.pointclouds.org code repository from March 2011 until December 2011. The code base is represented as a graph with edges that represent directories, and nodes that represent files. The icons jumping around from one part of the graph to another represent PCL developers. The colors of each node are just meant to represent the type of information stored (C++ source file, documentation, etc).

The video exhibits perfectly the collaborative nature of the project with almost continuous 24/7 changes and code commits. PCL's success draws precisely from this: assembling a pool of smart and creative international developers, that come together with complementary skills, ideologies, and different working schedules. :)

For more videos, please visit our YouTube channel at http://www.youtube.com/user/pointclouds.

The following list of presentations describes certain aspects or modules in PCL, and have been assembled by former research interns at Willow Garage.

• Ryohei Ueda's presentation on Tracking 3D objects with Point Cloud Library (more details)
• Jochen Sprickerhof's presentation on Large Scale 3D Point Cloud Mapping in PCL (more details)
• Aitor Aldoma's presentation on Clustered Viewpoint Feature Histograms (CVFH) for object recognition (more details)
• Julius Kammerl's presentation on Point Cloud Compression (more details)
• Dirk Holz's presentation on the PCL registration framework (pre 1.0) (more details)
• Rosie Li's presentation on surface reconstruction (more details)
• Kai Wurm's presentation on 3D mapping with octrees (more details)

Last updated November 10, 2014

To facilitate the open testing of algorithms the PCL community recognizes the need for free to use data. As such a collection of open data sets can be found on the PCL SourceForge data repository, as well as the PCL GitHub data repository. The files in this repository are stored in the PCD format, which can be easily read using PCL tools.

In addition to the PCL repository there are many other data repositories containing free downloadable data (in various formats). The following is a list of such sites containing links to data that is free to use for educational work. We encourage everyone to please read each sites' licensing information for any data used and appropriately acknowledge the authors in any publications or derived work.

• 3D Point Clouds

  • Autodesk Research - Digital 210 King
    Sample scan and image data of the Autodesk Research office.
  • Autonomous Systems Lab (ETH)
    Laser and Kinect data from structured and unstructured environments. This group has a list of useful related links as well as a similar set of links to data.
  • Canadian Planetary Emulation Terrain 3D Mapping Dataset
    Laser scans of unstructured terrains from rover platforms. Includes Matlab code for data parsing.
  • Cornell RGB-D Dataset
    A collection of office and home interior data sets. Individual captures along with stitched scenes are available.
  • Labeled 3D Point Clouds
    Fully labeled point clouds of urban scenes around the CMU campus.
  • Leica Geosystems (projected data)
    Example databases of laser scan data.
  • libLAS Samples
    Sample data sets in the las and laz formats.
  • Pointools
    Sample data sets for pointools evaluation.
  • Pseudo-Simulated LiDAR
    Pseudo-simulated lidar data created for change detection experiments.
  • Robotic 3D Scan Repository
    Registered 3D scans of indoor and outdoor environments. Some data sets include thermal and color information.
  • SpecTIR Hyperspectral Airborne Rochester Experiment
    A collection of imagery using an airborne hyperspectral imager, a laser detection and ranging (LiDAR), and a high resolution multi-spectral imaging system over the Rochester, New York, area July 26 - 29, 2010.

• Localization and Mapping

  • Barcelona Robot Lab Dataset
    Contains over 12GB of data and is intended for use in mobile robotics and computer vision research. The area surveyed in this data set covers 10000 sq m of the UPC Nord Campus in Barcelona, and includes 3D point clouds, onboard robot imagery as well as imagery from a camera sensor network. Ground truth data is also provided.
  • KITTI Vision Benchmark Suite
    Mono and stereo camera data, including calibration, odometry and more.
  • MIT DARPA Grand Challenge Team
    Camera, lidar and log files, including source code for parsing logs.
  • Mobile Robot Programming Toolkit Datasets
    A few pages of RGB+D, 6D ground truth, image, laser and other data for indoor and outdoor scenes.
  • Navigation Assistance for the Visually Impaired, NAVI (University of Zaragoza)
    The group provides varied datasets containing RGB-D data with the goal to establish a benchmark for the evaluation of systems performing NAVI (Navigation Assistance for the Visually Impaired) based on RGB-D devices. The datasets contain color and depth data generated using an Asus XtionPro Live camera in different indoor scenarios. Calibration data and information is included.
  • Navlab SLAMMOT Dataset
    Contains data collected from the Navlab11 testbed. There are three loops through an urban environment with SICK, pose and visual data.
  • NAVVIS Indoor Dataset
    LIDAR scans of the interiors of the corridors of the TU Munich main campus, acquired slice-by-slice by a mapping trolley and aggregated into a big point cloud. The site also includes registered imagery, camera calibration, occupancy grid maps and more.
  • NetBot Laboratory Data Sets (Texas A&M University)
    Typical office corridor scenarios recorded with a Samsung Galaxy S3 i747M cell phone camera, including ground truth and camera intrinsics. Also, another data set with similar imagery captured using a Nikon 5100 mounted on a PackBot, again with ground truth.
  • The New College Vision and Laser Data Set
    Stereo and omni-directional imagery, lidar and pose data.
  • PeRL: Ford Campus Vision and Lidar Dataset
    Camera and lidar data, including source code and more.
  • Radish: The Robotics Data Set Repository
    Odometry, laser and sonar data, including log files and generated maps.
  • The Rawseeds Project
    High-quality multisensor datasets with ground truth built for the purpose of benchmarking robotic systems.
  • SLAM-related robotics data sets (University of Freiburg)
    Compilation of 2D maps in the form of raw and correct log files.
  • ViDRILO: The Visual and Depth Robot Indoor Localization with Objects information dataset
    ViDRILO is a dataset containing five sequences of annotated RGB-D images acquired with a mobile robot in two office buildings under challenging lighting conditions.

• Models

  • The Digital Michelangelo Project
    Data is free but must be ordered as it is quite large.
  • Large Geometric Models Archive (Georgia Tech)
    Collection of standard models.
  • Stanford 3D Scanning Repository
    Collection of standard models (bunny, dragon, etc) including range data and zippered reconstructions.

• Aerial Point Clouds (including DEM and similar)

  • ASTER, JPL
    Global DEM.
  • Crater Lake National Park DEMs
    Twelve 7.5-minute quadrangle maps covering the park. The data set includes 10m DEM in multiple formats with associated metadata.
  • GNet
    Large aerial lidar data set.
  • ISPRS Test On Extracting DEMs From Point Clouds
    First and last pulse laser data from overhead scans of urban and natural environments.
  • LidarData.com
    Large database of aerial lidar. Data sets must be ordered and may not be free.
  • Miljøministeriet Geodatastyrelsen
    Large database of aerial lidar of Denmark. Select "(DHM) Danmarks Højdemodel" under Korttype (Denmark Elevation Model under Map Type) to view the available surfaces, level sets and point clouds. Registration is free and data can be downloaded using registered accounts.
  • National Land Survey of Finland
    Elevation and similar data of Finland.
  • NOAA Coastal Services Center Data
    Coastal elevations, orthoimagery, and various other data. LiDAR data comes in various formats including LAS.
  • North Carolina Floodplain Mapping Information System
    LiDAR DEM 50, DEM 20 and bare earth data for counties of North Carolina.
  • OpenTopography
    Aerial lidar data.
  • Point Cloud Classification (Lomonosov Moscow State university)
    Aerial lidar data with class labeling.
  • USGS DEMs, geocomm
    Data sets must be ordered and may not be free.
  • USGS Historic DEMs
    Mount St. Helens DEMs before and after 1980 eruption. Possibly other free data sets.

• Face Scans

  • Cyberware Head & Face Color 3D Scanner
    Sample head and face data including original data, merged ply files and uv/color data.

• Volume Data (MRI and similar)

  • The Chapel Hill Volume Rendering Test Data Set
    Head, knee and cadaver MRI data.

• Object Detection

  • B3DO: Berkeley 3D Object Dataset
    Indoor object detection data sets.
  • CMU IKEA Kitchen Object Dataset
    A set of 432 images from 9 objects along with image captures of videos taken chest level in kitchen environments.
  • IRML Summer School Data Repository
    Contains links to models (3DS and PLY) of common indoor objects along with labeled point clouds (PCD) created from several view points.
  • NYU Depth Datasets
    Indoor object detection data sets.
  • PaCMan Object Database
    Database of objects from IKEA for use in the PaCMan project. The database includes object scans along with meshes.
  • RGB-D Object Dataset
    Indoor object detection data sets with many common household objects.
  • Semantic Database of 3D Objects (Technische Universität München)
    Stereo camera and lidar scans of typical household objects.

• Misc/Variety

  • Andreas Geiger, Perceiving Systems, Max Planck Institute for Intelligent Systems
    A large collection of links to a variety of data sets and software.
  • Commercial storage for LiDAR data
    This group helps GIS users and companies to find or publish LIDAR data and geospatial information.
  • Computer Vision Group (Technische Universität München)
    Kinect RGB-D data for visual SLAM, multiview 3D reconstruction data and other image based data sets.
  • Virtual Terrain Project
    3D LiDAR data, topographic and other data. This site also has many useful links.

• Localization and Mapping

  • Mobile Robot Programming Toolkit
    C++ library that includes many SLAM vaiants as well as tools for camera calibration.
  • OpenSLAM.org
    Several C++ and Matlab variants of SLAM and related tools.

• Mesh Processing

  • TransforMesh, MeshHOG and MVViewer
    C++ source for removal of self intersection as well as interest point detction in triangular meshes. Also, a multi-camera and mesh viewer is provided.

• Nearest Neighbors, Distance and Similar

  • libNabo
    A fast k nearset neighbor library (C++, python) for low-dimensional spaces. See also ethzasl_icp_mapping at ROS.org.
  • PQP - A Proximity Query Package
    A library for performing several types of proximity queries on geometric models.
  • STANN
    The Simple, Thread-safe Approximate Nearest Neighbor (STANN) C++ Library.

• Readers/Writers

  • libLAS
    C++ library for readinga nd writing las files.
  • LAS reader for Matlab
    C++ mex function for loading .las files into Matlab.

• Registration

  • Dual-Bootstrap ICP (binaries)
    Executables for Dual Bootsrap-ICP and Retinal Vessel Centerline Tracing. Also, test data and ppt files are provided.
  • libPointMatcher
    Modular ICP library (C++) with fast performances, ROS wrapper, example code, documentation and unit tests. See also ethzasl_icp_mapping at ROS.org.

• Shape and Appearance Editing

  • Pointshop3D
    Pointshop3D - Pointshop3D is a system for interactive shape and appearance editing of 3D point-sampled geometry.

• Shape Detection and Recognition

  • 3D SURF
    Matlab demo of spatio-temporal SURF for the description of 3D shapes.
  • FVS: A Content-Based Retrieval Library
    Cross-platform 2D image and 3D object analysis library intended for use within a content-based retrieval system.
  • IRML Summer School Software Repository
    Includes PCL-based sources for point cloud alignment, segmentation and clustering as well as some Kinect data processing tools.
  • SURFing the Point Clouds: Selective 3D Spatial Pyramids for Category-level Object Recognition
    Paper, video, results and source for category-level object recognition.

• Surface Extraction

  • Direct computing of surface curvatures for point-set surfaces
    Matlab implementation of the paper by Pinghai Yang and Xiaoping Qian.
  • Moving Least Squares Surface Library
    C++ implementation by Bruno Madeira of the methods described in "Defining Point-Set Surfaces by Nina Amenta adn Yong Joo Kil.

• Toolkits and Misc.

  • 3DTK — The 3D Toolkit
    3DTK — The 3D Toolkit provides algorithms and methods to process 3D point clouds. It includes automatic high-accurate registration (6D simultaneous localization and mapping, 6D SLAM) and other tools, e.g., a fast 3D viewer, plane extraction software, etc.
  • CGAL
    Computation Geometry Algorithms Library. A C++ library containing algorithms for processing point clouds, meshes and much more.
  • DielmoOpenLiDAR
    DielmoOpenLiDAR is an open source software with GNU GPL license based in gvSIG for the management of LiDAR data. It allows access, visualization and analysis of the original LiDAR data.
  • Geometric Tools
    C++ library for real-time computer graphics, physics, mathematics, geometry, numerical analysis, image analysis and more.
  • Taglut
    Topological and Gepometrical Library - a Useful Toolkit. This inlcudes C++ source (with doxygen) for manipulating meshes, point clouds and other classical data from algorithmic geometry.
  • Vision and Mobile Robotics Laboratory
    Includes a Mesh Toolbox for manipulating triangular surface meshes (smoothing, registering, etc). Also, source is available for object detection, modeling and tracking as well as stereo-lidar fusion.