We propose a deep learning based framework that learns data-driven temporal priors to perform 3D human pose estimation from six body worn Magnetic Inertial Measurement units sensors. Our work estimates 3D human pose with associated uncertainty from sparse body worn sensors. We derive and implement a 3D angle representation that eliminates yaw angle (or magnetometer dependence) and sho... https://www.deliandiver.com/product-category/dog-food-gently-cooked-2/