Neural-network-based Gesture Detection for Capacitive Sensing

This work considers the algorithms, which make it possible to control various devices with movements of our hands. Signals from a small number of capacitive sensors are used to detect the simplest gestures. The classical feedforward neural network supplemented by some properties of the spiking neural networks allows identifying the type of a gesture – the direction of a hand movement above the plane of sensors – based on sensor signals change. The results of numerical experiments are presented, which confirm the efficiency of the proposed gesture detection algorithm for controlling various devices.