BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Where Ideas Take Shapes - Yann Savoye\, Robert Gordon University
DTSTART:20190606T131500Z
DTEND:20190606T140000Z
UID:TALK124435@talks.cam.ac.uk
CONTACT:Rafal Mantiuk
DESCRIPTION:In this talk\, I will provide a summary of high-level ideas be
 hind the most significant research we made over the last decade in the are
 a of Computer Graphics\, Computer Vision\, Digital Fabrication and Deep Le
 arning.\n\nFirst\, a video-based graphics pipeline to capture surfaces in 
 motion is presented. In the context of full-body human performance digital
 ization\, dynamic surfaces are governed by complex non-rigid behaviors har
 d to parametrize. The key idea is to rely on dimension reduction and varia
 tional principles. In practice\, a low-dimensional subspace coupled with g
 eneralized barycentric coordinates is used to manipulate space-time deform
 able geometry. This space-based deformer helps to register template-based 
 deformation driven by multi-view silhouettes.\n\nNext\, a hollowing optimi
 zation algorithm based on the concept of honeycomb-cells structure for add
 itive manufacturing will be showcased. Nowadays\, 3D printing is an ubiqui
 tous technology enabling low-cost and fast prototyping for digital fabrica
 tion of real-world objects. The key idea is to infer a stress-based Vorono
 i tessellation suitable for carving harmonic porous in the solid.\n\nFinal
 ly\, a few-shot Generative Adversarial Network architecture is discussed t
 o perform deep learning tasks with less amount of training data. The key i
 dea is that the generation improves the classification while the classific
 ation improves the generation. Consequently\, the model can learn adaptive
 ly by switching between full supervised learning and semi-supervised learn
 ing. This modified GAN is less restricted in term of input and output.\n\n
 In closing\, the importance of optimality at the heart of computational mo
 dels for visual computing will be discussed. Also\, some potential directi
 ons to enrich our field and to improve existing techniques for Capturing\,
  Learning and Printing will be sketched.
LOCATION:SS03 Meeting Room\, Computer Laboratory
END:VEVENT
END:VCALENDAR