Electrostatic Sensing-Based Human Activity Recognition

Addressing issues such as compromised privacy, blind spots, strict environmental requirements, and high sensor costs in non-contact human activity recognition, this paper proposes a method that uses electrostatic induction to capture human activity. The method involves the analysis and processing of voltage signals induced by human activities via electrostatic induction, and achieves automatic recognition and monitoring of human actions using a 1D-CNN algorithm. The proposed method was tested in natural indoor environments, employing an 8 cm electrode to recognize human activity within a range of 1.5 meters. Experimental results demonstrate an identification accuracy of 80.8%, validating the effectiveness of the proposed approach.