Hybrid 3D Human Pose Estimation with Monocular Video and Sparse IMUs

arXiv:2404.17837v1 Announce Type: new
Abstract: Temporal 3D human pose estimation from monocular videos is a challenging task in human-centered computer vision due to the depth ambiguity of 2D-to-3D lifting. To improve accuracy and address occlusion issues, inertial sensor has been introduced to provide complementary source of information. However, it remains challenging to integrate heterogeneous sensor data for producing physically rational 3D human poses. In this paper, we propose a novel framework, Real-time Optimization and Fusion (RTOF), to address this issue. …

2d-to-3d abstract accuracy arxiv computer computer vision cs.cv cs.hc however human hybrid information sensor temporal type video videos vision