Running Exercise IMU Dataset with 20 Participants and Seven Drills

147 JSON files contain raw accelerometer and gyroscope data from four IMUs worn on wrists and ankles. Twenty healthy participants performed seven running drills, including carioca, high-knee running, and regular running, for 10 seconds each under supervision. Alexander Josef Müller recorded the data between July and September 2020 under the supervision of Philipp Niklas Müller.

Use Cases

• Classifying specific running drills based on multi-sensor IMU data from wrists and ankles.
• Training models for wearable-based fitness activity recognition using accelerometer and gyroscope signals.
• Analyzing biomechanical patterns of different running exercises from synchronized limb sensor data.
• Benchmarking time-series classification algorithms on a labeled, multi-participant motion dataset.

Strengths

• Data includes 147 recording files from 20 participants performing seven distinct exercises.
• Sensor configuration is specified: accelerometer range +/-16 g, gyroscope range +/-2000 deg/s, and data rate 52 Hz.
• Each exercise recording is standardized to 10 seconds and was performed under supervision.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The participant demographic is skewed (16 male, 4 female) and age range is limited (16-31 years).
• Row count is unknown, which may limit suitability assessment for large-scale model training.

Provenance

Source

figshare

Collection Method

Supervised recordings using four IMUs worn on the right hand, left hand, right foot, and left foot.

Time Range

July to September 2020

Freshness

Last updated 2026-05-17 14:15:39; freshness should be verified.