Robust Biometrics from Motion Wearable Sensors Using a D-vector Approach
- PDF / 949,435 Bytes
- 17 Pages / 439.37 x 666.142 pts Page_size
- 32 Downloads / 171 Views
Robust Biometrics from Motion Wearable Sensors Using a D-vector Approach Manuel Gil-Martín1 · Rubén San-Segundo1 José Manuel Pardo1
· Ricardo de Córdoba1
·
Accepted: 28 August 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract This paper proposes a d-vector approach for extracting robust biometrics from inertial signals recorded with wearable sensors. The d-vector approach generates identity representations using a deep learning architecture composed of Convolutional Neural Networks. This architecture includes two convolutional layers for learning features from the inertial signal spectrum. These layers were pretrained using data from 154 subjects. After that, additional fully connected layers were attached to perform user identification and verification, considering 36 new subjects. This paper compares the proposed d-vector approach with previous proposed algorithms using in-the-wild recordings in different scenarios. The results demonstrated the robustness of the proposed d-vector approach for in-the-wild conditions: 97.69% and 94.16% accuracies (for user identification) and 99.89% and 99.67% Areas Under the Curve (for user verification) were obtained using one (walking) or several activities (walking, jogging and stairs) respectively. These results were also verified in laboratory conditions improving the performance reported in previous works. All the analyses were carried out using public datasets recorded at the Wireless Sensor Data Mining laboratory. Keywords Person identification and verification · D-vector approach · Gait recognition · Motion wearable sensors
1 Introduction During last years, computer-based systems have exploited multi-sensor networks to model human behaviors in different applications. Two of these applications are focused on biometrics-based access management: person identification and verification. On the one hand, person identification consists in a 1-to-n matching task where individual biometrics are compared to a database of possible identities to recognize the person. This application is used when police scan fingerprints in border controls for people identification. On the other hand,
B 1
Manuel Gil-Martín [email protected] Speech Technology Group, Information Processing and Telecomunications Center, E.T.S.I. Telecomunicación, Universidad Politécnica de Madrid, Ciudad Universitaria, SN, 28040 Madrid, Spain
123
M. Gil-Martín et al.
verification is a 1-to-1 matching task where it is necessary to determine if the user is who he/she says he/she is. For instance, a verification application is used when a person unlocks his/her smartphone with a PIN-code spoken by the user, verifying that the speaker is who he/she says. Traditional biometric systems usually have a fixed location for controlling the access to located resources such as a computer system or a border control. These systems decide based on one shot and do not supervise the subject identity along time. However, in this rapidly evolving digital world, person identification and ver
Data Loading...
