Rotation-Invariant Gait Identification with Quaternion Convolutional Neural Networks

TL;DR

This paper introduces Quaternion CNN, a neural network architecture that is inherently rotation-invariant, significantly improving gait identification robustness across different device orientations compared to traditional CNNs.

Contribution

The paper presents Quaternion CNN, a novel architecture that achieves rotation invariance in gait classification and offers a new method for visualizing learned features.

Findings

Quaternion CNN outperforms traditional CNNs in rotation-invariant gait classification

The learned kernels can be visualized as basis-independent trajectory fragments

The approach enhances robustness to device orientation variations

Abstract

A desireable property of accelerometric gait-based identification systems is robustness to new device orientations presented by users during testing but unseen during the training phase. However, traditional Convolutional neural networks (CNNs) used in these systems compensate poorly for such transformations. In this paper, we target this problem by introducing Quaternion CNN, a network architecture which is intrinsically layer-wise equivariant and globally invariant under 3D rotations of an array of input vectors. We show empirically that this network indeed significantly outperforms a traditional CNN in a multi-user rotation-invariant gait classification setting .Lastly, we demonstrate how the kernels learned by this QCNN can also be visualized as basis-independent but origin- and chirality-dependent trajectory fragments in the euclidean space, thus yielding a novel mode of feature visualization and extraction.