PORTLAND, Ore. — Innovative software solutions abounded at the IEEE's 9th annual Symposium on 3D User Interfaces (3DUi-2014) where five research groups competed in its 3D User Interface Contest. The winner, for the fourth time in the last five years, was the Center for Human-Computer Interaction at Virginia Polytechnic Institute and State University (Virginia Tech, Blacksburg, Va.).
Sponsored by the IEEE Computer Society the contest this year was to create a fast, natural-to-learn, and easy-to-use way of annotating 3D point clouds -- dense sets of data points in 3D space typically acquired by advanced 3D laser scanners. Point clouds are increasingly being used to automatically create detailed 3D geometric models. However, to be useful, the different components of point clouds need to be identified by a human and annotated according to their functions.
For instance, a 3D point cloud of the interior of a building might be generated by a laser scanner during a walk-through of the building. Then a human goes through the 3D dataset identifying rooms, equipment, and other assets in the building. The contest entailed creating a user-interface for adding such annotations that was fast, natural-to-learn, and easy-to-use, yet allowed for precise identification and labeling of its components. For the contest, a 3D point cloud of an insect was provided to contestants by the contest managers, with the task to annotate where its antennas, legs, feelers, and other components were located by adding textual labels to highlighted regions of the point cloud.
Winning Virginia Tech team members (left to right standing) are Cristian Moral Martos of Madrid, Spain; Mahdi Nabiyouni of Tehran, Iran; Doug Bowman, faculty adviser and professor of computer science at Virginia Tech. Seated is Felipe Bacim of Porto Alegre, Brazil.
(Source: Virginia Tech)
Contestants from Virginia Tech, Escola Politecnica da Universidade de São Paulo (Brazil), Universidade Federal do Rio Grande do Sul (Brazil), University of Würzburg (Germany), and University of Strasbourg (France) used widely varying approaches to the 3D point-cloud annotation problem. All groups wrote custom software driven by such hardware devices as a 3D mouse, a Leap Motion free-air gesture controller, virtual reality head-mounted displays, the PrimeSense 3D depth camera (now owned by Apple), and a novel combo 2D touchscreen and 3D pointer.
Virginia Tech took first place using a 3Dconnextion Spacepilot Pro 3D mouse combined with the Leap Motion free-air gesture controller and its Slice-and-Swipe software.
Next Page: Overcoming Leap's limitations