There has been significant prior work on learning realistic, articulated, 3D statistical shape models of the human body.
In contrast, there are few such models for animals, despite their many applications in biology, neuroscience, agriculture, and entertainment.
The main challenge is that animals are much less cooperative subjects than humans: the best human body models are learned from thousands of 3D scans of people in specific poses, which is infeasible with live animals.
In the talk I will illustrate how we extend a state-of-the-art articulated 3D human body model (SMPL) to animals learning from toys a multi-family shape space that can represent lions, cats, dogs, horses, cows and hippos.
The generalization of the model is illustrated by fitting it to images of real animals, where it captures realistic animal shapes, even for new species not seen in training.
Biography: Silvia Zuffi graduated in Electronic Engineering at the University of Bologna (Italy). After graduation she worked for some time in industry, where she contributed to implementing a system for tide forecast in Venice. Then, she joined Istituti Ortopedici Rizzoli, a orthopaedic hospital in Bologna (Italy), where she developed a method for the estimation of knee prostheses kinematics.
She worked for several years at the Italian National Research Council (CNR), where her research interests was colour imaging, specifically multispectral imaging and reproduction, and readability of coloured text on Web pages.
In 2008 she won a Marie Curie fellowship, and then started a PhD in computer vision at Brown University with the supervision of Micheal J. Black. She graduated with a thesis on Shape Models of the Human Body for Distributed Inference. In 2015 she was a postdoc at the MPI Institute for Intelligent Systems, in the Perceiving System department in Tuebingen, Germany. She is now a research scientist at the CNR Institute for Applied Mathematics and Information Technologies in Milan (Italy), where her current research interest is modeling the shape of animals for applications in computer vision and graphics.