ECVW2009

Fifth IEEE Workshop on Embedded Computer Vision

The Fifth IEEE Workshop on Embedded Computer Vision

(held in conjuction with ICCV 2009)
October 3rd, 2009

Contact: (ecv09 at computervisioncentral dot com)
http://www.computervisioncentral.com/content/ecvw09

Call for Participation (PDF).


AIT Awards for the best paper of Embedded Computer Vision 2009 are:

  • RACKET: Real-time Autonomous Computation of Kinematic Elements in Tennis, Julien Pansiot, Ahmed Elsaify, Benny Lo, Guang-Zhong Yang (Imperial College London, UK)
  • Real-Time Accurate Stereo with Bitwise Fast Voting on CUDA, Ke Zhang, Jiangbo Lu, Gauthier Lafruit, Rudy Lauwereins, Luc Van Gool (KU Leuven, Belgium)

Details and photos available on this news article.

Program (Click here to download PDF)

09h30 Welcome (Ahmed Nabil Belbachir, program chair)

S1: Keynote Session 1 (09h35 — 10h20)
Keynote: GPU – Accelerating Computer Vision
Khanh Vo Duc (NVIDIA)

(chair: Ahmed Nabil Belbachir, program chair)
Recently GPU has been emerging as an important means of taking parallelism to mainstream computing, from high-end supercomputer to low-end embedded devices. This keynote presentation provides an introduction to computer vision on the GPU. It tries to answer the fundamental question of why GPU for computer vision by covering the details of hardware and software aspects in using GPU to solve both traditional problems in computer vision, such as optical flow estimation, image stitching, and the modern ones as augmented reality, gesture estimation, vehicle collision detection, just to name a few. Finally, some challenges in harnessing the horsepower of GPUs for innovative embedded computer vision applications will be considered.

S2: Applications with Real-time Vision (10h20 — 11h00)
(chair: Abbes Amira, Brunel University)
10h20 RACKET: Real-time Autonomous Computation of Kinematic Elements in Tennis, Julien Pansiot, Ahmed Elsaify, Benny Lo, Guang-Zhong Yang (Imperial College London, UK)
10h40 An Embedded AER Dynamic Vision Sensor for Low-Latency Pole Balancing, Jorg Conradt, Raphael Berner, Matthew Cook, Tobi Delbruck (ETH Zürich, Switzerland)

11h00 Morning Break

S3: Real-time Stereo Vision (11h30 — 12h30)
(chair: Khanh Vo Duc, NVIDIA)
11h30 A Very Fast Census-based Stereo Matching Implementation on a Graphics Processing Unit, Michael Weber, Martin Humenberger, Wilfried Kubinger (AIT Austrian Institute of Technology, Austria)
11h50 Real-Time Accurate Stereo with Bitwise Fast Voting on CUDA, Ke Zhang, Jiangbo Lu, Gauthier Lafruit, Rudy Lauwereins, Luc Van Gool (KU Leuven, Belgium)
12h10 Geodesic Tree-Based Dynamic Programming for Fast Stereo Reconstruction, Chin-Hong Sin, Chia-Ming Cheng, Shang-Hong Lai, Shan-Yung Yang (NTHU, Taiwan)

12h30 Lunch Break

S4: Keynote Session 2 (14h00 — 14h45)
14h00 Keynote: Multiresolution Analysis for Computer Vision: From Algorithms to Low Power Reconfigurable Architectures, Abbes Amira (Brunel University, UK)

(chair: Ahmed Nabil Belbachir, AIT Austrian Institute of Technology)
This talk will focus on the software and hardware related issues for the design and implementation of multiresolution algorithms used in computer vision applications. A range of algorithms, including discrete wavelet transform (DWT), Ridgelet and Curvelet transforms will be presented for applications in medical image segmentation and compression and biometrics. The implementation process of these algorithms on Field Programmable Gate Arrays (FPGAs) will be described. This talk will review FPGA technology and methods for developing novel architectures for mutilresolution techniques used in computer vision applications, and will conclude with comprehensive case studies demonstrating the use of these low power efficient architectures to accelerate computationally-intensive image processing applications such as medical image segmentation and face.

S5: Poster Session (14h45 — 15h45)
(chair: Tobi Delbruck, ETH Zürich, Switzerland)
Efficient Keypoint Matching for Robot Vision using GPUs , Michael Schweitzer, Hans-J. Wünsche (UBM, Germany)
Probabilistic Model of Error in Fixed-Point Arithmetic Gaussian Pyramid , Antoine Méler, John Alexander Ruiz Hernandez, James L. Crowley (INRIA, France)
Efficient Systolic Architecture and Power Modeling for Finite Ridglet Transform , Abdul Sazish, Shrutisagar Chandrasekaran, Abbes Amira (Brunel University, UK)
MCMC-based Feature-guided Particle Filtering for Tracking Moving Objects from a Moving Platform, Chung-Ching Lin, Wayne Wolf (Georgia Tech, USA)
Flexible Clustering in Networks of Smart Cameras, Bernhard Dieber, Bernhard Rinner (Klagenfurt University, Austria), Nikolaus Viertl (AIT Austrian Institute of Technology, Austria)
A self-calibration method for smart video cameras, Roman Pflugfelder, Georg Nebehay (AIT Austrian Institute of Technology, Austria)
Towards Complex Visual Surveillance Algorithms on Smart Cameras Oliver Sidla (SLRE, Austria), Norbert Brändle (Arsenal Research, Austria), Wanda Benesova, Marcin Rosner, Yuriy Lypetskyy (Joanneum Research, Austria)
Fast Image Segmentation and Texture Feature Extraction for Image Retrieval, Tse-Wei Chen, Yi-Ling Chen, Shao-Yi Chien (National Taiwan University, Taiwan)
Parallel Algorithms to a Parallel Hardware: Designing Vision Algorithms for a GPU, Jun-Sik Kim, Myung Hwangbo, Takeo Kanade (CMU, USA)

15h45 Afternoon Break

S6: Real-time Detection & Tracking (16h00 — 17h50)
(chair: Roman Pflugfelder, AIT Austrian Institute of Technology)
16h00 GPU-based non-parametric background subtraction for a practical surveillance system, David Schreiber, Michael Rauter (AIT Austrian Institute of Technology, Austria)
16h20 An FPGA-Based Stream Processor for Embedded Real-time Vision with Convolutional Networks, Clément Farabet, Cyril Poulet, Yann LeCun (New York University, USA)
16h40 Realtime Affine-photometric KLT Feature Tracker on GPU in CUDA Framework, Jun-Sik Kim, Myung Hwangbo, Takeo Kanade (CMU, USA)
17h00 Hardware Architecture for High-Accuracy Real-Time Pedestrian Detection with CoHOG Features, Masayuki Hiromoto, Ryusuke Miyamoto (Kyoto University, Japan)
17h20 Parallel Implementation of pedestrian tracking using multiple cues on GPGPU, Hiroki Sugano, Ryusuke Miyamoto (Kyoto University, Japan)

17h40 AIT Award for the best paper of Embedded Computer Vision 2009 & Closing Remarks

Keynote Session 1(09h35 — 10h20)

Khanh Vo Duc is CUDA Technology Specialist at NVIDIA Japan, where he is working to support CUDA development teams at major companies and universities, who are creating innovative CUDA-based applications, from low-end image/video processing for consumer market, to signal processing for automotive manufactures, and data processing to high-end HPC customers. His first experience with CUDA was at Denso ITS Lab in implementing CUDA-enabled neural network for real-time object detection. Prior to CUDA, he has experiences using GPU for motion vector field analysis in video retrieval at Tokyo Institute of Technology. His research interests include visual object recognition, content-based image/video retrieval, and fundamental parallel algorithms on HPC platforms.

Keynote Session 2 (14h00 — 14h45)

Dr. Abbes Amira is a senior lecturer in Electronics and Computer Engineering in the School of Engineering and Design, Brunel University, UK (from May 2006). Previously, he held a lectureship in Computer Science at Queen’s University, Belfast, UK after receiving his Ph.D. in 2001 from the same university. He has been awarded a number of grants from government and industry, including EPSRC, Andor Technologies, TWI, Agility, NVIDIA and Inscentinel. He has published over 100 publications and successfully supervised 7 PhD students during his career to date. He was leading the medical imaging and embedded systems research at the Institute of Electronics, Communications and Information Technologies (ECIT), QUB Belfast, where he carried out successful research in the field of high performance reconfigurable computing and medical image analysis. He is now the leader of the image and vision systems, and reconfigurable computing research at Brunel University. He is a senior member of IEEE, senior member of ACM, Fellow of the Higher Education Academy and a member of IET. His research interests include: Image processing, Reconfigurable Computing, Embedded Systems for Image and Signal Processing, Low Power Design and SoC, Hardware Software CO-Design, Biometrics, Information Retrieval technologies and Medical Imaging.

General Chair: 
Nikos Bellas, University of ThessalyProgram Chair: 
Ahmed Nabil Belbachir, AIT Austrian Institute of Technology

Steering Committee: 
Sek Chai, Motorola Inc
Branislav Kisačanin, Texas Instruments
Jan-Michael Frahm, UNC Chapel Hill, USA 
Marc Pollefeys, ETH Zurich Switzerland, UNC Chapel Hill, USA 
Boaz J. Super, Motorola Inc.

Program Committee: 
Kristian Ambrosch, AIT Austrian Institute of Technology, Austria 
Nikolaos Bellas, University of Thessaly, Greece
Moshe Ben-Ezra, Siemens Corporate Research
Shuvra Bhattacharyya, University of Maryland
Terry Boult, University of Colorado at Colorado Springs
Sek Chai, Motorola Inc
Xin Chen, Navteq
James Coughlan, Smith-Kettlewell Institute
Goksel Dedeoglu, Texas Instruments
Ed Deprettere, Leiden University
Khanh Duc, NVidia 
Jan-Michael Frahm, UNC-Chapel Hill
Antonio Gentile, University of Palermo, Italy
Antonio Haro, DG FastChannel
Masatoshi Ishikawa, University of Tokyo
Branislav Kisačanin, Texas Instruments
Richard Kleihorst, NXP Research
Reinhard Koch, U. Kiel, Germany
Ajay Kumar, IIT Dehli, India
Zhu Li, Hong Kong Polytechnic University, Hong Kong
Ruei-Sung Lin, Google
Abelardo Lopez-Lagunas, ITESM-Toluca, Mexico
Jiebo Luo, Kodak
Yong Ma, Omron Corporation
Roberto Manduchi, University of California, Santa Cruz
Steve Mann, U. Toronto, Canada
Larry Matthies, Jet Propulsion Laboratory
Hongying Meng, University of York
Jeff Mulligan, NASA Ames
Vitorrio Murino, University of Verona, Italy
Rajesh Narasimha, Texas Instruments
Burak Ozer, Verificon Corporation
Marc Pollefeys, ETH Zurich Switzerland, UNC Chapel Hill, USA
Kari Pulli, Nokia Research Center
Bernhard Rinner, Klagenfurt University of Austria
Sezai Sablak, Bosch
Mainak Sen, Cisco Systems
Mubarak Shah, U. of Central Florida
Vinay Sharma, Texas Instruments
Yu Shi, NICTA, Australia
Joe Stam, Nvidia
Boaz J. Super, Motorola Inc
Salvatore Vitabile, University of Palermo, Italy
Linda Wills, Georgia Institute of Technology
Wayne Wolf, Georgia Institute of Technology
Feng Xiao, Fairchild Imaging
Xing Xie, Microsoft Research Asia
Ruigang Yang, U. Kentucky, USA
Tianli Yu, Like.com
Lin Zhong, Rice University

Previous Embedded Computer Vision Workshops (ECVW)