Improving Human Gait Recognition Using Feature Selection

Abstract

Abstract. Human gait, a biometric aimed to recognize individuals by the way they walk has recently come to play an increasingly important role in visual surveillance applications. Most of the existing approaches in this area, however, have mostly been evaluated without explicitly considering the most relevant gait features, which might have compromised the performance. In this paper, we have investigated the effect of discarding irrelevant or redundant gait features, by employing Genetic Algorithms (GAs) to select an optimal subset of features, on improving the performance of a gait recognition system. Experimental results on the CASIA dataset demonstrate that the proposed system achieves considerable gait recognition performance.

Applied Machine Learning on Phase of Gait Classification and Joint-Moment Regression

“…This work aims to be less biased and dependent on a specific body part (in the event there is limb loss), and instead demonstrates the numerical trade-offs between tracking different joints on the human body. To increase the efficiency of our model, we reduced the number of features used while training to reduce the training run-time and ease the process of data collection [19]. Our models used a maximum of n = 10 features to train: joint angles (θ 2D = 6), and four participant attributes.…”

“…The paper begins by explaining the process for preparing each data set using OpenSim, a multi-body biomechanic simulation package [5], and bioMechZoo, an open-source toolbox for analyzing and visualizing movement [6] to accurately label the joint angle coordinates and phase of gait in Section 2.1. To make sense of the aggregated data set, we implemented machine learning classification algorithms such as random forest [19,20] to correlate all of the joint angle recordings with a phase of gait [2] in Section 2.2. The analysis in Section 3 demonstrated that even with a large set of joints being tracked for a cyclical movement such as gait along the sagittal plane, there is no real requirement for wearing additional sensors.…”