A Convolutional Neural Network for Ultra-Wideband Radar-Based Hand Gesture Recognition
Sakorn Mekruksavanich, Ponnipa Jantawong, Datchakorn Tancharoen, Anuchit Jitpattanakul · 2023
New developments in sensing technology have enabled the creation of improved assistive devices that enhance daily eldercare routines and offer personalized care to users. Wearable or ambient sensors can now detect a person's behavior, but the need for constant recharging due to their high energy consumption makes wearing such devices around the clock difficult. In this paper, a new sensing technique based on deep learning (DL) approaches for hand gesture recognition (HGR) using an ultra-wideband (UWB) radar sensor is introduced. The study examines the recognition performance of a convolutional neural network (CNN) for HGR and utilizes the UWB-gestures dataset to train and test the DL networks in various experiments with different scenarios. The experimental results demonstrate that the CNN achieved the highest F1-score of 97.19%.