OPPORTUNITY: Wearable, Object and Ambient Sensor Activity Data

Overview

OPPORTUNITY combines wearable, object and ambient sensing for activity recognition. The UCI release contains four participants, each with five daily-living recordings and one scripted drill recording in a simulated studio flat.

Verified raw-archive structure

• All 24 measurement files were read: 869,387 rows, each with 250 whitespace-separated fields.
• Column 1 is time in milliseconds. Columns 2 to 243 contain 242 sensor attributes.
• The sensor attributes comprise 145 body-worn, 60 object and 37 ambient attributes.
• Columns 244 to 250 contain seven annotation fields: locomotion, high-level activity, left-arm action, left-arm object, right-arm action, right-arm object and combined-arm mid-level gesture.
• The archive documentation specifies synchronized data at 30 Hz.

Missing data and counting caveats

The 24 measurement files contain 8,380,350 literal NaN cells. Null activity annotations are a separate concept from missing sensor measurements. The UCI headline of 2,551 instances does not match the distributed measurement rows; its precise counting basis was not established. The row and column counts here describe the complete raw measurement tables, not a preprocessed challenge subset.

Use in experiments

Preserve participant and recording boundaries when forming evaluation splits. Any channel selection, missing-value treatment or removal of null labels changes the evaluated data and should be reported.

Source and citation

The UCI source record names Daniel Roggen, Alberto Calatroni, Long-Van Nguyen-Dinh, Ricardo Chavarriaga and Hesam Sagha. Its citation year is 2010, while its donation date is 8 June 2012. These dates serve different purposes.

Dataset DOI: 10.24432/C5M027. The bundled README requires acknowledgment through one of its two reference papers; one is The Opportunity challenge: A benchmark database for on-body sensor-based activity recognition. Notification by email is appreciated, rather than stated as mandatory.

The UCI record licenses the data under CC BY 4.0. Reuse requires appropriate attribution, a license link and an indication of changes. This catalog description is an independently prepared summary.