RPC-Net Dataset: High-Density Forearm EMG and Hand Kinematics

Simultaneous recordings of high-density surface electromyography (HD-sEMG) and hand kinematics from a motion capture system. The dataset was created by Giovanni Rolandino at the University of Oxford to support the development of the RPC-Net machine learning model for hand position estimation. It contains 96 monopolar EMG channels sampled at 2.048 kHz and 21 hand markers tracked at 100 Hz.

Use Cases

• Training neural networks for hand pose estimation based on forearm electromyography signals.
• Developing myoelectric control interfaces for prosthetics based on high-density EMG recordings.
• Analyzing the relationship between muscle activity and hand joint angles based on simultaneous EMG and motion capture data.
• Benchmarking algorithms for decoding motor intent from non-invasive biosignals.

Strengths

• High-density EMG with 96 channels provides detailed spatial muscle activity information.
• Synchronous acquisition of EMG and precise VICON motion capture data enables direct signal-to-pose mapping.
• EMG data was sampled at a high frequency of 2.048 kHz, capturing detailed signal dynamics.

Limitations

• Row count and dataset size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

University of Oxford

Collection Method

Data was acquired using a MEACS HD-sEMG system on the forearm and a VICON motion capture system with reflective markers on the hand.

Geography

Data acquisition location is unspecified, though the equipment was developed at Politecnico di Torino in Italy.