Reproduce “OESense: Employing Occlusion Effect for In-ear Human Sensing” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dong Ma, Andrea Ferlini, Cecilia Mascolo
OESense introduces an earable sensing system that leverages an inward-facing microphone and the natural occlusion effect to enable robust human motion detection. It achieves over 99% accuracy for step counting, approximately 98% for human activity recognition, and up to 97% for hand-to-face gesture recognition, even in the presence of head movements or environmental noise.
Key insights
- The occlusion effect within a sealed ear canal naturally amplifies low-frequency bone-conducted vibrations from human motions, improving signal strength.
- An inward-facing microphone combined with the occlusion effect intrinsically suppresses external environmental noise, leading to a higher signal-to-noise ratio.
- Motion-induced signals primarily reside in low-frequency bands (<50 Hz), allowing them to be effectively separated from music playback and enabling co-existence.
- Small amounts of user-specific data significantly improve the generalization and accuracy of activity and gesture recognition models across individuals.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.