Kickboxing Gesture IMU Recordings from 21 Participants

2,519 IMU recordings of eight full-body kickboxing-inspired gestures performed by 21 participants. The data includes recordings from four sensors placed on the wrists and ankles, configured with accelerometer and gyroscope ranges. It was recorded in July 2019 by Christian Müller under the supervision of Philipp Niklas Müller.

Use Cases

• Training machine learning models for mobile gesture recognition based on IMU data.
• Benchmarking human activity recognition algorithms on dynamic, full-body movements.
• Studying sensor fusion techniques using data from four synchronized IMU boards.
• Analyzing the biomechanics of martial arts gestures from wrist and ankle sensor perspectives.

Strengths

• Contains 2,519 recording files, providing a substantial sample size for model training.
• Includes data from 21 participants with documented demographics (age, handedness, experience).
• Specifies sensor configuration details, such as accelerometer range +/-16 g and gyroscope range +/-2000 deg/s.

Limitations

• Description metadata is limited; actual data quality and structure require manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred from the JSON files.
• The participant pool of 19 male and 2 female individuals may reflect a gender bias.

Provenance

Source

Recorded by Christian Müller under supervision of Philipp Niklas Müller.

Collection Method

Data was collected using four Silicon Labs Thunderboard Sense 2 boards per recording.

Time Range

July 2019

Freshness

Last updated 2026-05-17 14:28:35; freshness should be verified.