Smartphone-Based Gait Recognition Dataset: Naturalistic Walking Data from 390 Participants for Biometric Identification

This dataset enables realistic biometric identification research with naturalistic gait data from 390 participants. Smartphone inertial sensors (triaxial accelerometer, gyroscope, and magnetometer at 30 Hz) captured 46.8 kilometers of walking across approximately 585 minutes, with participants holding devices in their dominant hand. Multiple sessions per participant on different days ensure robust representation of intra-subject variability. Comprehensive metadata and documentation support deep learning model development, gait-based authentication systems, and bench marking of recognition algorithms under real-world conditions.