Computer Vision Systems

Computer vision research has taken great strides over the past decade. To - ploy researchresults in large-scale,real-worldapplications, progress on scienti?c frontiersmustbecomplementedbycorrespondingresearchanddevelopmentc- cerning issues such as systems integration, scalable architectures and repres- tations, automated con?guration, adaptation and recovery, closing percepti- action-loops, real-time and embedded implementations, benchmarking, etc. The InternationalConferenceonComputerVisionSystems(ICVS)isaregularforum with this perspective on systems. It brings together researchers and developers from academia and industry, fostering research and technology transfer relevant to real-world, deployed computer vision systems. This volume contains the papers presented at ICVS 2009 in Li` ege, Belgium, continuing the established series of conferences held so far in Europe and North America. A total of 96 submissions were received. Each of them was reviewed by at least 3 program committee members in a double-blind procedure. Overall, 45 contributed papers were accepted, 21 for oral presentation and 24 as posters. The contributions cover a broad spectrum of aspects related to computer - sion systems such as learning, recognition, HCI, robotic applications, real-time constraints, cognitive systems, and architectures, to name just a few. In addition, ICVS 2009 featured two invited speakers who presented c- plementary issues of high importance to computer vision systems. James M. DiCarlo from the Massachusetts Institute of Technology is a leading researcher incomputationalmodels ofbrainmechanismsthatunderlie biologicalvision,the ultimate example of deployedvisionsystems and animportant source of inspi- tionfor arti?cialsystems.JayYagnikistheHeadofComputer VisionandAudio Understanding Research at Google Inc., overseeing research and development in computer vision aimed at extremely large-scale application.