Coverage-Aware IMU Augmentation for Human Activity Recognition
September 16, 2026
A new augmentation framework for IMU-based wearable sensors uses learned sensor embedding spaces to identify data scarcity and generate targeted virtual candidates. This approach improves model generalization in low-resource settings by weighting virtual samples based on their coverage utility.
HOW THIS AFFECTS YOU
●
builderYou can improve the robustness of wearable sensor models when training data is limited.
●
healthThis could lead to more reliable continuous monitoring tools for diverse patient populations.