CVPR 2022 best paper candidate | pip: six inertial sensors realize whole body dynamic capture and force estimation

This paper presents a physical based sparse inertial dynamic capture and human force estimation scheme ：Physical Inertial Poser (PIP). Use only 6 An inertial sensor （Inertial Measurement Unit,IMU）, This scheme can capture human motion in real time, which conforms to the physical laws of the real world , Joint stress 、 As well as the ground force and other information .

The system can be found in CPU In order to 60fps Speed of operation , The algorithm delay is only 16 millisecond , Compared with previous work, it achieves the highest attitude estimation accuracy on public data and 、 Movement smoothness 、 And the lowest system delay , And the force estimation of human body based on sparse inertial sensor is realized for the first time . By introducing physical optimization , This scheme greatly improves the realism and accuracy of the estimated human motion . The study was selected CVPR 2022 Best Paper Finalist.

Project home page ：https://xinyu-yi.github.io/PIP/

Thesis link ：https://arxiv.org/abs/2203.08528

Open source code ：https://github.com/Xinyu-Yi/PIP