How Unique do we Move? : Understanding the Human Body and Context Factors for User Identification

Abdrabou, Yasmeen and Mecke, Lukas and Rivu, Radiah and Prange, Sarah and Nguyen, Quy Dat and Voigt, Vanessa and Alt, Florian and Pfeuffer, Ken (2023) How Unique do we Move? : Understanding the Human Body and Context Factors for User Identification. In: Proceedings of Mensch und Computer 2023 :. Mensch und Computer 2023 . ACM, CHE, pp. 127-137. ISBN 9798400707711

Text (_MuC_2023____Full_body_identification)
_MuC_2023_Full_body_identification.pdf - Accepted Version
Download (1MB)

Abstract

Past work showed great promise in biometric user identification and authentication through exploiting specific features of specific body parts. We investigate human motion across the whole body, to explore what parts of the body exhibit more unique movement patterns, and are more suitable to identify users in general. We collect and analyze full-body motion data across various activities (e.g., sitting, standing), handheld objects (uni- or bimanual), and tasks (e.g., watching TV or walking). Our analysis shows, e.g., that gait as a strong feature amplifies when carrying items, game activity elicits more unique behaviors than texting on a smartphone, and motion features are robust across body parts whereas posture features are more robust across tasks. Our work provides a holistic reference on how context affects human motion to identify us across a variety of factors, useful to inform researchers and practitioners of behavioral biometric systems on a large scale.