Support Vector Machine to Recognize Hand Motions Using Body Worn Flexible Antenna

Microwave signal based human activity recognition technique using Support Vector Machine (SVM) is proposed in the paper. The paper has employed a flexible patch antenna and exploited the impedance characteristics for the detection of different hand activities. The antenna is embedded on hand to capture the variation of the input impedance due to the hand motions. The experiment has considered six elderly subjects and attached the patch antennas to their wrists. The impedance data sets for six different hand activities are collected and analyzed for activity recognition. The Discrete Wavelet Transform (DWT) technique is used for extracting the features from the real and imaginary impedance data set. A supervised model is employed to analyze the original and the augmented data sets. Application of SVM on the raw data sets bestows 90.62% classification accuracy whereas using the support vector machine combined with data augmentation and DWT technique offers classification accuracy up to 95%. Moreover, the experiments reveal that out of six people two have a tremoring illness and two are suffering from slight tremors which may be an indication of a neurological disorder.