Improved Static Hand Gesture Classification on Deep Convolutional Neural Networks Using Novel Sterile Training Technique

In this paper, we investigate novel data collection and training techniques towards improving classification accuracy of non-moving (static) hand gestures using a convolutional neural network (CNN) frequency-modulated-continuous-wave (FMCW) millimeter-wave (mmWave) radars. Recently, non-contact pose static gesture recognition have received considerable attention in many applications ranging from human-computer interaction (HCI), augmented/virtual reality (AR/VR), even therapeutic range motion for medical applications. While most current solutions rely on optical or depth cameras, these methods require ideal lighting temperature conditions. mmWave radar devices recently emerged as promising alternative offering low-cost system-on-chip sensors whose output signals contain precise spatial information non-ideal imaging Additionally, deep networks been employed extensively image by learning both feature extraction simultaneously. However, little work has done radars CNNs due to the difficulty involved extracting meaningful features return signal, results are inferior compared with dynamic classification. This article presents an efficient approach technique CNN introducing ``sterile'' images which aid distinguishing distinct among subsequently improve accuracy. Applying proposed yields increase rate $85\%$ $93\%$ $90\%$ $95\%$ range-angle profiles, respectively.