The focus of this workshop is on presenting, discussing and demonstrating recent tracking methods and applications that work well in practice and that show some superiority over state-of-the-art methods. Rather than focusing on pure novelty, this workshop encourages presentations that concentrate on complete systems and integrated approaches.
The TMA workshop looks at pose tracking from an end-to-end point of view. Practical tracking solutions for AR are usually complex systems that combine many methods including, but not limited to:
Two keynote talks given by invited world-renowned specialists in fields from which Augmented Reality can highly benefit will be delivered at the beginning of the presentation sessions. The fields include computer vision, other sensing modalities (e.g. GPS, inertial, etc.) and systems issues (calibration, estimation, fusion, etc.).
Monday, November 5, 9:00 - 17:30
Georgia Tech Hotel and Conference Center
|09:15||Invited speaker: Andrew Davison|
|slides||How Increasing Frame-Rate Improves Real-Time Tracking|
|Higher frame-rates promise better tracking of rapid motion, but advanced real-time vision systems rarely exceed the standard 10-60Hz range, arguing that the computation required would be too great. Actually, increasing frame-rate is mitigated by reduced computational cost per frame in trackers which take advantage of prediction. Additionally, when we consider the physics of image formation, high frame-rate implies that the upper bound on shutter time is reduced, leading to less motion blur but more noise. So, putting these factors together, how are application-dependent performance requirements of accuracy, robustness and computational cost optimised as frame-rate varies? Using 3D camera tracking as our test problem, and analysing a fundamental dense whole image alignment approach, we open up a route to a systematic investigation via the careful synthesis of photorealistic video using ray-tracing of a detailed 3D scene, experimentally obtained photometric response and noise models, and rapid camera motions. Our multi-frame-rate, multi-resolution, multi-light-level dataset is based on tens of thousands of hours of CPU rendering time. Our experiments lead to quantitative conclusions about frame-rate selection and highlight the crucial role of full consideration of physical image formation in pushing tracking performance.|
|Single Camera Calibration using partially visible calibration objects based on Random Dots Marker Tracking Algorithm|
|paper||A mobile markerless Augmented Reality system for the automotive field|
|Towards practical inside-out head tracking for mobile seating bucks|
|Virtualized reality model-based benchmarking of AR/MR camera tracking methods|
|14:00||Invited speaker: Jan-Michael Frahm|
|Fast and Scalable Crowd Sourced Image Registration and Dense Reconstruction of the World|
|In recent years photo and video sharing web sites like Flickr and YouTube have become increasingly popular. Nowadays, every day millions of photos are uploaded. These photos survey the world. Given the scale of data we are facing significant challenges to process them within a short time frame given limited resources. In my talk I will present our work on the highly efficient organization and reconstruction of 3D models from city scale photo collections (millions of images per city) on a single PC in the span of a day. The approaches address a variety of the current challenges to achieve a concurrent 3D model from these data. For registration and reconstruction from photo collections these challenges are: selecting the data of interest from the noisy datasets, and efficient robust camera motion estimation. Shared challenges of photo collection based 3D modeling and 3D reconstruction from video are: high performance stereo estimation from multiple views, as well as image based location recognition for topology detection. In the talk I will discuss the details of our image organization method, our efficient stereo fusion technique for determining the scene depths from photo collection images.|
|slides||Hardware and Software Trends in Mobile AR|
|Participants will form small groups to separately discuss advances/ limitations/requirements of topics such as RGB-D camera-based tracking, tracking in outdoor environment, parallel/simultaneous tracking and reconstruction, tracking on mobile devices, sensor fusion, etc.|
|16:15||Reports and discussions for breakouts|
|One representative of each break-out group will report on the discussion|
|Demos from presenters and participants|
|17:30||End of the workshop|
Participation is open to all attendees registering for the workshop, not only to presenters. We explicitly invite non-research attendees to join us as well and provide valuable insight into commercial areas of interest and constraints.
Andrew Davison, Imperial College London
Andrew Davison received a BA in physics and the D.Phil. degree in computer vision from the University of Oxford in 1994 and 1998, respectively. He undertook his doctorate with with Prof. David Murray at Oxford's Robotics Research Group, where he developed one of the first robot simultaneous localisation and mapping (SLAM) systems using vision. He then spent two fantastic years as a European Union (EU) Science and Technology Fellow at the National Institute of Advanced Industrial Science and Technology (AIST), Japan, where he continued to work on visual robot navigation. In 2000 he returned to the University of Oxford as a Postdoctoral Researcher working with Ian Reid and was awarded a five-year Engineering and Physical Sciences Research Council (EPSRC) Advanced Research Fellowship in 2002. During this time he developed the well known MonoSLAM algorithm for real-time SLAM with a single camera. He joined Imperial College London as a lecturer in 2005, where he teaches robotics in the Department of Computing and leads the Robot Vision Research Group. In 2008, he was awarded a five year European Research Council (ERC) Starting Grant. The group's work continues to focus on the challenges in real- time, real-world 3D vision, expanding on the core problems of localisation and mapping with cameras towards a more general real-time model-based scene understanding agenda.
The wide applicability of this research in robotics and beyond into areas like augmented reality, gaming, mobile devices and automotive has been proven by strong industrial interest and the group has ongoing links with companies in several different sectors. Recent work has been recognized with best paper awards at ICRA 2010 and ISMAR 2011, and the best demonstration award at ICCV 2011.
Jan-Michael Frahm, University of North Carolina at Chapel Hill
Jan-Michael Frahm is an Assistant Professor at University of North Carolina at Chapel Hill. He received his Ph.D in computer vision in 2005 from the Christian-Albrechts University of Kiel, Germany. His Diploma in Computer Science is from the University of Lübeck. Jan-Michael Frahm’s research interests include a variety of computer vision in the areas of structure from motion, dense 3D reconstruction, location recognition and the development of data-parallel algorithms for commodity graphics hardware for efficient 3D reconstruction. He has written more than 70 peer-reviewed publications in computer vision conferences and journals and is an editor in chief of the Elsevier Journal of Image and Vision Computing.