GAITEX: Human motion dataset of impaired gait and rehabilitation exercises using inertial and optical sensors

Overview

GAITEX is a high-fidelity human motion dataset introduced in Nature Scientific Data. It addresses the need for detailed datasets covering physiotherapeutic exercises and gait disorders, providing synchronized recordings from wearable sensors and reference optical systems.

Data Collection and Instrumentation

• The dataset captures participants performing normal gait on treadmills and specific rehabilitation exercises under controlled conditions.
• Instrumentation includes nine body-worn Inertial Measurement Units (IMUs) capturing raw orientation data and 35 to 68 optical markers for full-body kinematic reference.
• A unique feature is the mounting of optical markers directly onto IMU housings, allowing for precise technical validation of sensor-derived orientations against ground-truth motion capture.

Dataset Contents

• Raw orientation data from nine IMUs and time-synchronized Motion Capture (MoCap) data.
• Processed versions including IMU orientations aligned to segment coordinate systems and subject-specific OpenSim models.
• Manual curated temporal annotations delineating individual repetitions of therapeutic tasks.
• Movement quality ratings for rehabilitation exercise assessment.

Use Cases

• Development and validation of IMU-only methods for human movement analysis.
• Training machine learning models for exercise quality evaluation and gait classification.
• Temporal segmentation of physiotherapeutic activities and biomechanical parameter estimation.

Access and License

The dataset is published in Nature Scientific Data and is available for research use. Users should consult the publisher page for exact licensing terms and proper citation instructions.