Learning based representation has become the key to the success of many
computer vision systems. While many 3D representations have been proposed, it
is still an unaddressed problem how to represent a dynamically changing 3D
object. In this paper, we introduce a compositional representation for 4D
captures, i.e. a deforming 3D object over a temporal span, that disentangles
shape, initial state, and motion respectively. Each component is represented by
a latent code via a trained encoder. To model the motion, a neural Ordinary
Differential Equation (ODE) is trained to update the initial state conditioned
on the learned motion code, and a decoder takes the shape code and the updated
state code to reconstruct the 3D model at each time stamp. To this end, we
propose an Identity Exchange Training (IET) strategy to encourage the network
to learn effectively decoupling each component. Extensive experiments
demonstrate that the proposed method outperforms existing state-of-the-art deep
learning based methods on 4D reconstruction, and significantly improves on
various tasks, including motion transfer and completion.

本研究提出一种基于深度学习的方法，通过编码器将每个组件表示为潜在编码，使用神经 ODE 对初始状态进行更新，使用解码器将形状编码和更新状态编码组合在一起重构 3D 模型，并通过身份交换训练策略来有效地解耦每个组件，从而在 4D 重建及运动转移任务中取得了显著改进。