REACT: A Conditioning Framework for User-Adaptive sEMG Hand Pose Estimation
For sEMG-based hand pose estimation, REACT offers a practical solution to inter-user variability without retraining, enabling personalized inference on wearable devices.
REACT personalizes a frozen pretrained sEMG-to-pose model for unseen users using a lightweight conditioning framework with FiLM, reducing angular error by up to 3.9% on the EMG2POSE benchmark with under 45 seconds of calibration.
Surface electromyography (sEMG) enables continuous hand pose estimation on wearable devices, but models trained on multi-user corpora degrade on unseen individuals due to inter-user variability in anatomy and electrode placement. We propose REACT, a lightweight conditioning framework that personalizes a frozen pretrained EMG-to-pose backbone at inference time using only a handful of calibration recordings. REACT learns a compact user embedding from calibration data and applies Feature-wise Linear Modulation (FiLM) to adapt the shared encoder's feature space, requiring no gradient updates at deployment. On the large-scale EMG2POSE benchmark, REACT improves over the state-of-the-art baseline across all three generalization splits in both regression and tracking modes, reducing angular error by up to 3.9% with minimal parameter overhead and under 45 seconds of per-user calibration.