scene-aware global human motion forecasting is critical for manifold
applications, including virtual reality, robotics, and sports. The task
combines human trajectory and pose forecasting within the provided scene
context, which represents a significant challenge.
So far, only Mao et a