Speed-Up 3D Human Pose Estimation Task Using Sub-spacing Approach

Abstract

This paper tackles the problem of reconstructing 3D human poses from given 2D landmarks, which is still an ill-posed problem. The existing works have successfully applied Active Shape Model approach to estimate 3D human poses, but the execution time is quite high. In this paper, we propose a speed-up method by separating data into subspaces to reduce the execution time of existing methods in two steps: (i) Predicting the subspace that the need-estimated 3D shape could belong to. (ii) Estimating 3D shape from given 2D landmarks and predefined basis shapes of this subspace. Compare to existing works; our approach shows a significant reduction in computational time.