Gait-based Person Identification using Multiple Inertial Sensors

Inertial sensors such as accelerometers and gyroscopes have gained popularity in recent years for their use in human activity recognition. However, little work has been done on using these sensors for gait-based person identification. Gait-based person identification turns out to be important in applications such as where different people share the same wearable device and it is desirable to identify who is using the device at a given time while walking. In this research, we present the first multi-sensory gait-based person identification dataset EJUST-GINR-1 and present our work on gait-based person identification using multi-sensory data, by mounting 8 wearable inertial sensory devices on different body locations and use this data to identify the person using it. Two of these sensors are smart watches worn on both wrists. We explore the correlation between each body location and the identification accuracy, as well as exploring the effect of fusing pairs of sensory units in different locations, on the final classification performance.