Robust Inertial Motion Tracking through Deep Sensor Fusion across Smart Earbuds and Smartphone
summary
A system called DeepIT fuses phone and earbud IMUs (using Nokia eSense earbuds streaming gyro/accel over BLE) with an LSTM reliability network and attention-based fusion plus a reliability-weighted magnetometer compensation to fix angular drift. A new large-scale smartphone+earbud inertial dataset covers four walking patterns (NW, RP, RPE, HM). Built end-to-end, trained on that data, and evaluated against closed-form and data-driven baselines, it yields multiple-fold accuracy improvements in uncontrolled natural walking scenarios.
AI-written summary · about
related talks
13mUltraPoser: Pushing the Limits of IMU-based Full-Body Pose Estimation with Ultrasound Sensing on ...
26mMaking Sense The Road to mobile Awareness
27mDissecting ActivityRecognition with Multi-Sensor Simulation and Its Practical Deployment
22mRunning, walking, sitting or biking? - Motion prediction with acceleration and rotation
11mBaroPoser: Real-time Human Motion Tracking from IMUs and Barometers in Everyday Devices
07mDecentralized Attention-based Personalized Human Mobility Prediction
10mHAGI: Head-Assisted Gaze Imputation for Mobile Eye Trackers
14mDeepFisheye: Near-Surface Multi-Finger Tracking Technology Using Fisheye Camera