TL;DR本文介绍了一种使用动力学模型来辅助深度学习神经网络进行关节式物体姿态估计的方法,并在玩具和人类 3D 姿势问题的数据实验中证明了该方法的有效性。
Abstract
Learning articulated object pose is inherently difficult because the pose is
high dimensional but has many structural constraints. Most existing work do not
model such constraints and does not guarantee the geome