Person Identification based on Smartphones Inertial Sensors
Rafael Saraiva Campos, Lisandro Lovisolo · 2018
Several works explore the use of measurements from smartphones embedded inertial sensors (INS) to recognize which physical activity the person who is carrying the device is executing, in what is generally called Human Activity Recognition (HAR). In this work, we propose an extension to HAR, using INS readings also to identify specific subjects. The individual inertial “signature” of a person is expected to be more distinguishable while walking than in quasi-static states (laying, sitting, standing), so we restrict our proposal to personal identification of walkers. We refer to this feature as Activity Recognition with Person Identification (AR-PID). To implement it, we use supervised ensemble learning with bagging (bootstrap aggregating), defining majority voting committees of multilayer perceptrons (MLPs) and training them to perform as specialized binary classifiers for each user. To evaluate the feasibility of AR-PID, we use a public domain INS database, with 10299 samples collected by 30 smartphone users. A classification accuracy higher than 90% is achieved for 86.7% of the users. We also investigate the use of subsequent sets of measures to improve identification performance.