Radar-Based Hand Gesture Recognition Using Spiking Neural Networks

We propose a spiking neural network (SNN) approach for radar-based hand gesture recognition (HGR), using frequency modulated continuous wave (FMCW) millimeter-wave radar. After pre-processing the range-Doppler or micro-Doppler radar signal, we use signal-to-spike conversion scheme that encodes Doppler maps into spike trains. The trains are fed recurrent network, liquid state machine (LSM). readout signal from SNN is then used as input different classifiers comparison, including logistic regression, random forest, and support vector (SVM). Using machines of less than 1000 neurons, achieve better state-of-the-art results on two publicly available reference datasets, reaching over 98% accuracy 10-fold cross-validation both data sets.