Timezone: »
Variational Autoencoder with Differentiable Physics Engine for Human Gait Analysis and Synthesis
Naoya Takeishi · Alexandros Kalousis
Event URL: https://openreview.net/forum?id=9ISlKio3Bt »
We address the task of learning generative models of human gait. As gait motion always follows the physical laws, a generative model should also produce outputs that comply with the physical laws, particularly rigid body dynamics with contact and friction. We propose a deep generative model combined with a differentiable physics engine, which outputs physically plausible signals by construction. The proposed model is also equipped with a policy network conditioned on each sample. We show an example of the application of such a model to style transfer of gait.
Author Information
Naoya Takeishi (HES-SO / RIKEN)
Alexandros Kalousis
Related Events (a corresponding poster, oral, or spotlight)
-
2021 : Variational Autoencoder with Differentiable Physics Engine for Human Gait Analysis and Synthesis »
Dates n/a. Room
More from the Same Authors
-
2021 Poster: Learning interaction rules from multi-animal trajectories via augmented behavioral models »
Keisuke Fujii · Naoya Takeishi · Kazushi Tsutsui · Emyo Fujioka · Nozomi Nishiumi · Ryoya Tanaka · Mika Fukushiro · Kaoru Ide · Hiroyoshi Kohno · Ken Yoda · Susumu Takahashi · Shizuko Hiryu · Yoshinobu Kawahara -
2021 Poster: Physics-Integrated Variational Autoencoders for Robust and Interpretable Generative Modeling »
Naoya Takeishi · Alexandros Kalousis -
2015 Poster: Space-Time Local Embeddings »
Ke SUN · Jun Wang · Alexandros Kalousis · Stephane Marchand-Maillet