Walker-Independent Features for Gait Recognition from Motion Capture Data

TL;DR

This paper introduces walker-independent gait features derived from raw motion capture data, enabling effective human identification even with limited labeled data and new identities appearing on the fly.

Contribution

It proposes a novel modification of Fisher Linear Discriminant Analysis with Maximum Margin Criterion to learn gait features that generalize across different walkers.

Findings

Features can discriminate unseen individuals.

Effective with fewer training identities than encountered walkers.

Supports real-time identification in dynamic environments.

Abstract

MoCap-based human identification, as a pattern recognition discipline, can be optimized using a machine learning approach. Yet in some applications such as video surveillance new identities can appear on the fly and labeled data for all encountered people may not always be available. This work introduces the concept of learning walker-independent gait features directly from raw joint coordinates by a modification of the Fisher Linear Discriminant Analysis with Maximum Margin Criterion. Our new approach shows not only that these features can discriminate different people than who they are learned on, but also that the number of learning identities can be much smaller than the number of walkers encountered in the real operation.