Sorry, you need to enable JavaScript to visit this website.

Style-Driven Multi-Resolution Human Motion Synthesis from Limited Data

Citation Author(s):
David Eduardo Moreno-Villamarin, Anna Hilsmann, Peter Eisert
Submitted by:
David Moreno-Vi...
Last updated:
8 February 2024 - 6:00am
Document Type:
Example videos
David Eduardo Moreno-Villamarin

We present a generative model that learns to synthesize human motion from limited training sequences. In contrast to existing methods, our framework provides stylistic control across multiple temporal resolutions. The model adeptly captures human motion patterns by integrating skeletal convolution layers and a multi-scale architecture. Our framework contains a set generative and adversarial networks, along with style embedding modules, each tailored for generating motions at specific frame rates while exerting control over their style. Through direct synthesis of SMPL pose parameters, our approach avoids test-time adjustments to fit human body meshes. Experimental results showcase our model's ability to achieve extensive coverage of training examples, while generating diverse motions, as indicated by local and global diversity metrics.

0 users have voted: