Keynote Speakers

Reinhard Koch - Depth Cameras

Reinhard Koch graduated from the University of Hannover, Germany, and received his PhD (Dr.-Ing.) on the topic of 3D scene reconstruction from stereoscopic image sequences in 1996. From 1996-99 he worked with Prof. Luc van Gool in Leuven, Belgium, as head of the 3D reconstruction group. Since 1999 he is professor of Computer Science at the Christian-Albrechts-University Kiel, Germany. His research field is multimedia information processing, with emphasis on 3D scene reconstruction from images and video. He is author and coauthor of over 150 peer reviewed articles at international conferences, journals, and book chapters, associate editor of two international journals in computer vision, and programm committee member of numerous international conferences. He currently serves as speaker of the technical committee of the German Association for Pattern Recognition, DAGM, and is German representative at the Governing Board of the International Association on Pattern Recognition, IAPR. His current research interests are underwater 3D reconstruction, high-quality multi view scene capture, autostereoscopic 3D-TV, and deformable 3D scene analysis from range video sequences.

The analysis and reconstruction of 3D scenes from video data is an important theme in computer vision. Recently, novel cameras have emerged that allow to capture not only color, but also the depth of the scene in realtime. Such devices, like LIDAR, time-of-flight imaging, or the Kinect game sensor from Microsoft, facilitate 3D scene analysis enormously and allow for a complete range of novel applications. In my presentation I will first explain the technique and ability of such depth cameras, their combination with traditional color cameras, and then discuss some recent application areas, like auto-stereoscopic 3D-Television, capture and tracking of deformable moving objects, or virtual studio 3D graphics applications, all of which rely on the novel depth devices.

Dave Ferguson - Self-Driving Cars

Dave Ferguson is a researcher working on Google's self-driving car program. His current research focuses on machine learning, planning, and perception in the context of autonomous vehicles. His algorithms have been used by a number of real-world robotic systems including the Mars Exploration Rovers, subterranean mine mapping robots, driverless cars, and robot manipulators. From 2006 - 2007 Dave was the planning lead for "Boss", Carnegie Mellon University's winning entry in the DARPA Urban Challenge. From 2006 - 2008 Dave was the co-lead of the Personal Robotics project at Intel Research, where he helped lead the development of "Herb", an autonomous mobile manipulation robot for indoor assistance. He received his BSc(Hons) degree in Maths and Computer Science from Otago University in 2001, and his Ph.D. in Robotics from Carnegie Mellon University in 2006.

Most of us use a car every day. But unlike airplanes, which have been flying on autopilot for decades, cars are still driven manually - just the way they were driven 100 years ago. This talk will discuss the Google Self-Driving Car Project, an attempt to develop passenger vehicles that are able to drive themselves. Building on early research in the 90's in Germany and the US, and more recently the DARPA Challenges, driverless car technology has now advanced to be within reach of commercial application. I'll describe some of the underlying artificial intelligence approaches that have enabled Google's fleet of self-driving cars to navigate more than 180,000 miles on public roads in California and Nevada, including the downtowns of San Francisco and Los Angeles.

Nick Barnes - Computer Vision for Prosthetic Vision

Nick Barnes is a principal researcher and leader of the computer vision research group at NICTA, and is a lead investigator of the Bionic Vision Australia consortium which aims to develop a bionic eye. He has been at NICTA's Canberra Research Laboratory since 2003. He was a lecturer at the University of Melbourne from 2000 to 2003, a visiting researcher at the LIRA Lab, University of Genoa, Italy in 1999. He completed his BSc (Hons) and PhD in computer vision for robot navigation at the University of Melbourne, in 1999. He has authored or co-authored more than 90 publications in refereed journals and conferences. His research interests focus on computer vision and prosthetic vision.

Bionic Vision Australia is a consortium of Australian research organisations working to develop a retinal implant to restore sight. This talk describes work at NICTA to develop vision processing for implantable prosthetic vision. I describe the key challenges, along with the role of computer vision in implantable prosthetic vision, and approaches for addressing these challenges. Our approaches use computer vision to derive information about particular objects in the scene that are important to task performance. We then use this understanding of the scene to ensure that key visual information is preserved through augmentation of the resulting prosthetic visual image. We show in human trials using simulated prosthetic vision that such an approach can yield benefits for an orientation and mobility task.

Important Dates

  • 17th September - Extended deadline for subissions (5-6 pages) for review.
  • 10th September 2012 - Deadline for submissions (5-6 pages) for review.
  • 15th October 2012 - Notification of acceptance.
  • 29th October 2012 - Final papers due.
  • 26th-28th November - Conference dates.

The conference proceedings are now available through the ACM Digital Library.

IVCNZ Dunedin 2012
University of Otago Google ACM ICPS