Presentation
Text-based Motion Synthesis With a Hierarchical Two-stream RNN
SessionSimulation & Animation
Event Type
Poster
Artificial Intelligence/Machine Learning
Deep Learning
Modeling
Robotics
Research & Education
Ultimate Supporter
Ultimate Attendee
Exhibitor Ultimate
Enhanced Attendee
Time
Location
DescriptionWe propose an RNN-based sequential network to map natural language text descriptions to 3D pose sequences for human motions where the input texts may describe single actions with sequential information, e.g., ”a person walks four steps forward“ or multiple superimposed actions, e.g., “a person is spinning around while walking.”
The authors of this Poster have been invited to participate in the Technical Papers Summary and Q&A: Motion Capture, Tuesday, 10 August, 2-3 pm.
The authors of this Poster have been invited to participate in the Technical Papers Summary and Q&A: Motion Capture, Tuesday, 10 August, 2-3 pm.
