Biometric Database for Human Gait Recognition using Wearable Sensors and a Smartphone

This paper presents a multi-model biometric authentication database for human gait recognition. In contrast to the most known database available in literature that uses either machine vision, floor sensor based or wearable sensor techniques, we developed a public multi-model biometric database for human gait using wearable sensors and smartphone. Human gait data characteristics were recorded for 50 subjects (37 males and 13 females) of age range varying from 14 to 52 years old. Experimental gait data recording was collected using five wearable shimmer sensor modules that were attached to various locations on human body) in addition to a Samsung Galaxy Note Smartphone (with built-in accelerometer and gyroscope sensors) held in hand. Different walking scenarios like slow, normal and fast walk, in addition to multiple co-variables such as age, weight, and height were investigated. Equal Error Rate (EER) in different walking scenarios ranged from 0.17% to 2.27% for the five wearable sensors at different locations, where as EER results of smartphone data ranged from 1.23% to 4.07%. Average Genuine Reject Rate (GRR) of sensors located at the leg, pocket and hand decreased with the increase of age group, while it did not follow any trend for sensors located at the upper pocket and in the bag. Moreover GRR results on all sensors show no significance regarding height or weight variations.