Sandro Schönborn received a Master of Science in Physics from the University of
Basel in 2008. He studied different optimization methods for chemical structure
prediction and focused on computational physics. Beginning in 2009, he switched
to computer vision and joined the group of Prof. Thomas Vetter, where he studies
image interpretation with generative models. For integrative face image
analysis, he received his PhD in computer science from the University of Basel
in 2013 and now continues his work as a postdoctoral researcher. Sandro
Schönborn also regularly teaches robotics at a local high school.
I am interested in how machines can see and understand images. At a more
specific level, my main area of research lies with generative models of face
images where my daily work revolves around the 3D Morphable Model.
I mainly study how to use and adapt the model to interpret, manipulate and
synthesize face images. During my PhD studies, my main focus lied on
integrative model fitting methods. The methods integrate useful but unreliable
low-level image information, contrary to standard top-down model fitting
methods. During this work, I developed a strong interest in Data-Driven Markov
Chain Monte Carlo Methods as a formalization of a propose-and-verify strategy.
Such a concept offers a powerful way to deal with unreliable data by splitting
the model fitting process into a "proposal" and a "verification" stage.
I am also interested in quantitative analysis of face perception and the
associated manipulation of face images. Using a statistical and generative
model, it becomes possible to change face images to appear differently to human
observers, for example more trustworthy.