3D human motion prediction

current project :)

Predicting how human motion evolves in the short-term future is a task that has many applications in human-computer interaction, computer vision, or robotics and is an important building block in systems used for autonomous navigation (e.g. self-driving cars). Human motion data can come in many forms. In this project, we are working with skeletal representations, i.e. we represent the human body as a set of its major joints which is organized in a tree-like, hierarchical chain that defines how the joints are linked together. Each joint is parameterized as a rotation that defines how one bone segment must be rotated to obtain the orientation of the following bone segment.

We are given sequences of pre-recorded human motion data and our task is to predict how the motion continues for several frames in the future.

Code coming after we are finished with the project :D