Introducing a Human Activity Recognition Dataset Gathered on Real-Life Conditions

Abstract

[Abstract] Human activity recognition (HAR) has garnered significant scientific interest in recent years. The widespread use of smartphones enabled convenient and cost-effective data collection, eliminating the need for additional wearables. Given that, this paper introduces a novel HAR dataset in which participants had freedom in choosing smartphone orientation and placement during activities, ensuring data variability. It also includes contributions from diverse individuals, reflecting unique smartphone usage habits. Moreover, it comprises measurements from accelerometer, gyroscope, magnetometer, and GPS, corresponding to one of four activities: inactive, active, walking, or driving. Unlike other datasets, the collected data in this study were obtained from smartphones used in real-life scenarios