Abstract
The function of constructing the hierarchy of objects is important to the visual process of the human brain. Previous studies have successfully adopted capsule networks to decompose the digits and faces into parts in an unsupervised manner to investigate the similar perception mechanism of neural networks. However, their descriptions are restricted to the 2D space, limiting their capacities to imitate the intrinsic 3D perception ability of humans. In this paper, we propose an Inverse Graphics Capsule Network (IGC-Net) to learn the hierarchical 3D face representations from large-scale unlabeled images. The core of IGC-Net is a new type of capsule, named graphics capsule, which represents 3D primitives with interpretable parameters in computer graphics (CG), including depth, albedo, and 3D pose. Specifically, IGC-Net first decomposes the objects into a set of semantic-consistent part-level descriptions and then assembles them into object-level descriptions to build the hierarchy. The learned graphics capsules reveal how the neural networks, oriented at visual perception, understand faces as a hierarchy of 3D models. Besides, the discovered parts can be deployed to the unsupervised face segmentation task to evaluate the semantic consistency of our method. Moreover, the part-level descriptions with explicit physical meanings provide insight into the face analysis that originally runs in a black box, such as the importance of shape and texture for face recognition. Experiments on CelebA, BP4D, and Multi-PIE demonstrate the characteristics of our IGC-Net.
Abstract (translated)
构建对象层级的功能对于人类大脑的视觉过程非常重要。以往的研究成功采用了胶囊网络在不 unsupervised 的情况下将数字和人脸分解成部分来研究神经网络相似的感知机制。然而,他们的描述局限于 2D 空间,限制了他们模仿人类固有的 3D 感知能力的能力。在本文中,我们提出了一种Inverse Graphics Capsule Network (IGC-Net)来从大规模未标记图像中学习Hierarchical 3D 面部表示。IGC-Net的核心是一种新的胶囊,名为 graphics capsule,它在计算机图形中表示具有可解释参数的3D基本点,包括深度、反射率和3D姿态。具体来说,IGC-Net首先将对象分解成一组语义 consistent 的部分级描述,然后将它们组成对象级描述来构建层级。学习到的 graphics capsule 揭示了神经网络以视觉感知为导向理解人脸作为3D模型层级的重要性。此外,发现的部分可以用于 unsupervised 面部分割任务来评估我们方法的语义一致性。此外,具有明确物理意义的部分级描述提供了从原本运行在黑盒中的面容分析中得出的启示,例如形状和纹理对于人脸识别的重要性。在CelebA、BP4D 和 Multi-PIE 的实验中证明了我们的 IGC-Net 的特征。
URL
https://arxiv.org/abs/2303.10896