Capture-24: Human Activity Recognition from Wearable Sensor Data

Capture-24 is an activity tracker dataset for human activity recognition, superseded by a newer version. The dataset was created by researchers at the University of Oxford and is associated with studies published in Sage Journals and Nature Scientific Reports. It was used to test self-report time-use diaries against objective instruments and for statistical machine learning of sleep and physical activity phenotypes.

Use Cases

• Training activity classification models based on sensor data streams.
• Validating self-reported time-use diaries against objective sensor measurements.
• Identifying sleep and physical activity behavioral phenotypes from wearable device data.

Strengths

• Dataset is associated with peer-reviewed publications in Sage Journals and Nature Scientific Reports.
• Data originates from a major research institution, the University of Oxford.

Limitations

• This is an old version of the dataset that has been superseded by a newer release.
• Row count, column definitions, and file formats are unknown, limiting suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

University of Oxford

Collection Method

Likely collected via wearable activity trackers, as referenced in associated research papers.