A Comparative Study of Matrix Completion and Recovery Techniques for Human Pose Estimation

We present a comparative study of three matrix completion and recovery techniques, applied to the problem of human pose estimation. Human pose estimation algorithms may exhibit estimation noise or may completely fail to provide estimates for some joints. A post-process is often employed to recover the missing joints' locations from the available ones, typically by enforcing kinematic constraints or by using a prior learned from a database of natural poses. Matrix completion and recovery techniques fall into the latter category and operate by filling-in missing entries of a matrix, with the available/non-missing entries being potentially corrupted by noise. We compare the performance of three such techniques in terms of the estimation error of their output as well as their runtime under varying parameters. We conclude by recommending use cases for each of the compared techniques.