人間の拡張2D骨格関節点座標に基づくカメラ視点変化にロバストな行動認識

抄録

In recent years, with the rapid growth of computing power and data transmission capacity of portable devices, research on action recognition has attracted much attention. Cross-view action recognition is one of the important branches. Cross-view action recognition has always been a difficult problem for conventional traditional action recognition based on 2D data, also limits its practical application ability. To solve this problem, we propose a method to extend the 2D data into 3D, and then perform action recognition based on 3D data.Through a series of experiments, we find that the proposed method has robustness in the action recognition with camera viewpoint change．