May 6, 09:15-10:15
Andrew Zisserman, University of Oxford, United Kingdom
Human Pose Estimation and Retrieval in Videos
Determining the pose and activities of humans is one of the central problems of image and video analysis. The visual problem is challenging because humans are articulated animals, wear loose and varying clothing, self-occlude themselves, and stand against difficult and confusing backgrounds. Nevertheless, the area has seen great progress over the last decade due to advances in modelling and in the efficiency of algorithms.
We describe here recent progress in determining human 2D upper body pose in TV videos and films. The pose specifies the spatial layout of head, torso, upper and lower arms and hands. A standard approach to this problem is to learn a pictorial structure model and fit it to the video frames. We will describe two variations on this approach which differ in their level of supervision and the strength of the applied model. The first model requires explicit supervision, but is then able to track human pose through hour long videos. The second model is weaker (it is tree structured and there is no occlusion modelling), but requires no supervision.
We show results of these models on various TV videos and feature films, and applications of the estimated pose to (i) learning the gestures of sign language, and (ii) to pose based video retrieval.
This is joint work with Patrick Buehler, Mark Everingham, Vitto Ferrari, Daniel Huttenlocher and Manuel Marin-Jimenez
May 7, 09:00-10:00
Fernando Pereira, Instituto Superior Técnico, Lisbon, Portugal
Video Compression: Still Evolution or Time for Revolution?
Video compression has been steadily evolving in the past 20 years with significant developments in the technology, major compression gains, impressive deployment of services and applications, and an amazing overall impact on our society. Until recently, the 'video compression adventure' has been providing around 50% compression gains every 5 years which led to the set of video coding standards nowadays available. This set of standards resulted from a rather smooth technological evolution, with the same predictive coding architecture being successively enriched with new tools to provide higher compression factors, typically also at the cost of a higher encoding complexity.
However, something seems to have changed recently since, after the development of the very successful H.264/AVC standard, further compression gains have been short and more difficult to reach than usual. This fact led many video coding research experts to announce the end of the predictive video coding saga as known from the past two decades.
In this context, this talk will precisely discuss the future of video compression considering the emerging industry needs, notably in terms of 3D video and ultra high resolutions, promising technological novelties, and recent standardization initiatives. With MPEG and VCEG, the two main video coding standardization players running to announce new initiatives - High-Efficiency Video Coding (HVC) from MPEG and Next Generation Video Coding (NGVC) from VCEG - it is time to carefully analyze the possible ways to go forward in video compression and answer the big question: Is it still possible to extend the predictive video coding saga to get further significant compression gains or is it finally time for a revolution in terms of video coding concepts, architectures and tools ?
Fernando Pereira is currently with the Electrical and Computers Engineering Department of Instituto Superior Técnico (IST) and with Instituto de Telecomunicaçõ, both in Lisbon, Portugal. He is responsible for the participation of IST in many national and international research projects. He acts often as project evaluator and auditor for various organizations. He is an Area Editor of the Signal Processing: Image Communication Journal, a member of the Editorial Board of the Signal Processing Magazine, and is or has been an Associate Editor of IEEE Transactions of Circuits and Systems for Video Technology, IEEE Transactions on Image Processing, IEEE Transactions on Multimedia, and IEEE Signal Processing Magazine. He is a member of the IEEE Signal Processing Society Technical Committees on Image, Video and Multidimensional Signal Processing, and Multimedia Signal Processing, and of the IEEE Circuits and Systems Society Technical Committees on Visual Signal Processing and Communications, and Multimedia Systems and Applications. He was an IEEE Distinguished Lecturer in 2005 and an IEEE Fellow in 2008. He has been a member of the Scientific and Program Committees of many international conferences and has contributed more than 200 papers. He has been participating in the work of ISO/MPEG for many years, notably as the head of the Portuguese delegation, Chairman of the MPEG Requirements Group, and chairing many Ad Hoc Groups related to the MPEG-4 and MPEG-7 standards. His areas of interest are video analysis, processing, coding and description, and interactive multimedia services.