Pervasive Pose Estimation for Fall Detection

• DOI: 10.1145/3478027
• Publication Date: 2022-04-07T12:44:47Z
• Authors: Jiaqing Luo; Ruiyu Bai; Suining He; Kang G. Shin
• Keywords: Accidental; Barometer; Accidental fall; Mobile phone; Tilt sensor

Falls are the second leading cause of accidental or unintentional injuries/deaths worldwide. Accurate pose estimation using commodity mobile devices will help early detection and injury assessment falls, which essential for first aid elderly falls. By following definition fall, we propose a P ervasive ose Est imation scheme fall ( \( ^2 \) ), measures changes in tilt angle height human body. For measurement, leverages pointing device, e.g., smartphone, when unlocking to associate Device coordinate system with World system. exploits fact that person’s remains unchanged while walking calibrate pressure difference between device floor. We have prototyped tested various situations environments. Our extensive experimental results demonstrated can track body orientation irrespective pocket phone is placed in. More importantly, it enables phone’s barometer detect falls environments decimeter-level accuracy.