Recognizing Detailed Human Context In-the-Wild from Smartphones and Smartwatches

The ability to automatically recognize a person's behavioral context can\ncontribute to health monitoring, aging care and many other domains. Validating\ncontext recognition in-the-wild is crucial to promote practical applications\nthat work in real-life settings. We collected over 300k minutes of sensor data\nwith context labels from 60 subjects. Unlike previous studies, our subjects\nused their own personal phone, in any way that was convenient to them, and\nengaged in their routine in their natural environments. Unscripted behavior and\nunconstrained phone usage resulted in situations that are harder to recognize.\nWe demonstrate how fusion of multi-modal sensors is important for resolving\nsuch cases. We present a baseline system, and encourage researchers to use our\npublic dataset to compare methods and improve context recognition in-the-wild.\n