MSR Vision Faculty Summit – Adventures in the world of 3d Computer Vision

The unifying theme of my research is using visual data to extract mathematical representations of the 3d world. These representations aim to be able to faithfully reproduce aspects of the real world like its shape, appearance and motion. This talk will give a brief guided tour around some of the topics I have worked with in the past few years where the type of visual data used ranges from still images to video and from controlled studio to totally passive capture. Similarly the real world aspect modelled and type of application varies between static geometry for building graphics ‘assets’, to object deformations, with applications in face capture.

This is a still rapidly growing field; however these technologies are now showing levels of maturity that permit us for the first time to think about commercial-grade systems and solutions. I will also give my perspective on what I think are the key engineering challenges of putting these systems together, as well as what I believe should be the focus of research on this domain in the coming years.