Visual Computing Workshop, June 10—11, 2010
The Visual Computing Workshop is part of the Symposium on Visual Computing and Speech Processing, presented by TU Braunschweig's Center for Informatics and Information Technology (tubs.CITY). The workshop is supported by the Gesellschaft für Informatik, Fachbereich Graphische Datenverarbeitung (GI FB-GDV).
The Visual Computing Workshop will bring together experts from academia and industry to reflect on the state-of-the-art as well as on contemporary and upcoming challenges in visual computing research and applications. The workshop is aimed at junior and senior VC researchers and practitioners alike. Its goal is to strengthen the visual computing community and its profile in Germany, Europe, and beyond.
Date & Time
June 10-11, 2010
Thursday 2pm - Friday 12:30pm
Haus der Wissenschaft
38106 Braunschweig, Germany
The early-bird workshop registration fee is 130,-EUR.
Due to venue constraints, the number of participants is limited. Workshop registration takes place on a first come, first serve basis.
Thursday June 10, 2010
|12:30pm||Registration desk opens|
Multimodal Computing and Interaction - Robust, Efficient and Intelligent Processing of Text, Speech, Visual Data, and High Dimensional Representations
Hans-Peter Seidel, MPI InformatikThe past three decades have brought dramatic changes in the way we live and work. This phenomenon is widely characterized as the advent of the Information Society. Ten years ago, most digital content was textual. Today, it has expanded to include audio, video, and graphical data. The challenge is now to organize, understand, and search this multimodal information in a robust, efficient and intelligent way, and to create dependable systems that allow natural and intuitive multimodal interaction. The Cluster of Excellence on Multimodal Computing and Interaction, established by the German Research Foundation (DFG) within the framework of the German Excellence Initiative, addresses this challenge. The term multimodal describes the different kinds of information such as text, speech, images, video, graphics, and high-dimensional data, and the way it is perceived and communicated, particularly through vision, hearing, and human expression.
The GPU in Visual Computing
Timo Stich, NVidiaStarting from the first introduction of the GPU as a 3D graphics accelerator, it has come a long way. GPUs today are not only used for playing latest 3D games, but are a key technology that drives the Visual Computing revolution that we see today. This can be seen in the domains of image processing, video editing, compression, image analysis, and computational imaging. In this talk I will give an overview of the current hardware and software solutions that NVIDIA offers to produce and run Visual Computing applications.
Part-Based People and Object Detection
Bernt Schiele, TU Darmstadt & MPI InformatikAfter motivating part-based models for object class detection I will review our work on part based people and object detection. In the first part I will summarize our work that integrates people detection and tracking and that has recently been extend to articulated 3D pose estimation. In the second part of the talk I summarize two approaches for object class detection. The first uses probabilistic topic models to enable decomposition, discovery, and detection of visual object classes. The second is based on a shape-based model for object class detection that enables explicit knowledge transfer between object classes.
Applications of Free-Viewpoint Video in Broadcast
Oliver Grau, BBC Research & Development, LondonFree-viewpoint video has great potential for broadcast applications. Particular interest is for example in special effects to give programmes a 'fresh look' or for post-match analysis of sports events. This presentations gives an overview of some work on free-viewpoint video at BBC R&D; on studio applications and outdoor sport. Furthermore, some of the requirements and challenges for using these techniques in real productions are reviewed.
The World in the Browser: 3D-Internet using XML3D
Philipp Slusallek, Intel Visual Computing Institute & DFKI SaarbrückenThe Web supports a wide range of basic data and media types including text, audio, images, and now streaming video (HTML-5), which are increasingly used to create innovative applications in the browser. However, despite the ubiquitous availability of powerful graphics -- processors even in mobile devices -- interactive 3D graphics is not yet part of the Web. XML3D is minimal addition to HTML that allows to embed interactive 3D graphics into any HTML page while reusing the existing Web technology wherever possible (including DOM, events, JS/Scripting, CSS, AJAX, etc.). Its close integration into Web technologies, its modular design, and its hardware-friendly data types are also the main differences to VRML, X3D, and other technologies. Because XML3D objects are part of the DOM they can easily manipulated just like existing HTML content by any web programmer. The 3D-Internet promises to make interactive 3D graphics available to everyone but this creates a number of big challenges as well. In my talk I will discuss these challenges and opportunities -- and how they are addressed with XML3D. I will also present about a number of techniques and projects building on top of XMl3D.
|5:45pm||end of first day program|
|7:30pm||workshop dinner (restaurant Rodizio)|
Friday June 11, 2010
tba(keynote talk organized by co-located workshop on speech processing)
3D Video Processing
Peter Eisert, FhG-HHI & HU BerlinIn recent years, 3D video technology has enabled many new applications.3D cinema productions have become more and more common, and the progress in autostereoscopic displays make 3DTV also available to the home user. At the same time, high resolution camera technology has reached affordable prices and increased processing power even on mobile devices can handle these enormous data in real-time algorithms. In this talk, activities in the area of 3D video processing are presented, that focus on enabling such new applications. Multi-view processing techniques from camera arrays are shown and demonstrated for the applications of high resolution panoramic video and immersive 3D conferencing. In contrast to the regular video structure, 3D graphics exploited in virtual environments has different characteristics. As an example, we present a system for interactive streaming of computer graphics for remote gaming. Finally, both real video and synthetic 3D graphics are combined in augmented reality applications and a Virtual Mirror system for Virtual Try On of clothes and shoes will be presented.
Visual Computing for Multimodal Sensor Data
Andreas Kolb, Siegen University & speaker DFG-GK 1564Classically, Visual Computing is considered as the discipline of computer science which deals with the acquisition, representation, manipulation, analysis, synthesis and application of visual information, i.e. images and image sequences in a spatial and temporal context. An alternative definition is more user centered and defines Visual Computing as computing based on the interaction with and/or control work by manipulating visual images either as direct work objects or as objects representing other objects. This talk extends the first concept of Visual Computing by broadening the notion of "visual information" to a "data array acquired by sensors" (usually in 2D or 3D), where the sensors are not only classical imaging sensors, but also range sensors and sensors that possibly work at different wave length beside the visual one, e.g. THz- or multispectral infrared sensors. Regarding the second definition, this is still focused on "image objects", not arbitrary objects. The talk will give some examples of current research that fit to the given extended definition of Visual Computing. It will discuss the challenges of dealing with this kind of data also in case of multimodal applications.
3D Video and Free Viewpoint Processing and Coding
Aljoscha Smolic, Disney Research ZurichAn overview of 3D video and free viewpoint video is given with special focus on related standardization activities in MPEG. Free viewpoint video allows the user to freely navigate within real world visual scenes, as known from virtual worlds in computer graphics. 3D video provides the user with a 3D depth impression of the observed scene, which is also known as stereo video. In that sense as functionalities, 3D video and free viewpoint video are not mutually exclusive but can very well be combined in a single system. Research in this area combines computer graphics, computer vision and visual communications. It spans the whole media processing chain from capture to display and the design of systems has to take all parts into account. The conclusion is that the necessary technology including standard media formats for 3D video and free viewpoint video is available or will be available in the future, and that there is a clear demand from industry and user side for such new types of visual media.
Visual Computing in Diagnostic Medical Imaging
Hartmut Schirmacher, Visage Imaging GmbHDiagnostic medical imaging has been undergoing many rapid changes, especially since the introduction of fast multi-slice CT and MR scanners that produce 3D and 4D image data with sub-milimeter resolution. Although a vast body of techniques has been proposed and implemented for analyzing these images using automatic segmentation, quantification, and feature detection, today only very few of these so-called "advanced visualization" techniques can be found in the routine workflow of a typical radiology practice or community hospital. In this presentation I will review some interesting trends and features of the global diagnostic imaging IT and advanced visualization markets, and show examples of technologies and their actual or potential impact on radiology in hospitals and imaging centers. How can modern visual computing make a difference; what key requirements need to be met; and how can research and industry work together to really affect the future of radiology?