[empkins-members] EmpkinS Invited Talk - Marilyn Keller
On Thursday, 19 May 2022, we invite you to a talk by Marilyn Keller about "Inferring the anatomic skeleton of a subject from its external appearance". The talk starts at 2 pm.
Participation is obligatory and highly recommended for doctoral candidates.
Marilyn Keller is a Ph.D. student in the Department of Perceiving Systems at the Max Planck Institute for Intelligent Systems, advised by Professors Sergi Pujades and Michael J. Black. Her work focuses on building a statistical model of the anatomic human skeleton. Before, she got a Master of Science from Grenoble INP Phelma and Karlsruhe Institute. She also worked on 4D captures at INRIA Grenoble in the MORPHEO Team and at BBC R&D as part of internships. Her interests span Computer Vision, Computer Graphics, and Virtual Reality.
We address the problem of inferring the anatomic skeleton of a person, in an arbitrary pose, from the 3D surface of the body; i.e. we predict the inside (bones) from the outside (skin). Many existing state-of-the-art biomechanical skeletons are detailed but do not easily generalize to new subjects. Additionally, computer vision and graphics methods that predict skeletons are typically heuristic, not learned from data, and are not validated against ground truth.
In this work, we use 1000 male and 1000 female dual-energy X-ray absorptiometry (DXA) scans to build a dataset of paired body meshes and skeleton meshes. From this dataset, we learn the mapping from the 3D body surface to the internal skeleton. We name our inference tool OSSO, for "Obtaining Skeletal Shape from Outside". Given an arbitrary 3D body shape and pose, OSSO predicts a realistic skeleton inside.