Abstract

The proposed study provides a novel technique for recognizing hand gestures that use a combination of Deep Convolutional Neural Networks (DCNN) and 60 GHz Frequency Modulated Continuous Wave (FMCW) radar. The motion of a Human's hand is detected using the FMCW radar, and the various gestures are classified using the DCNN. Motion detection and frequency analysis are two techniques that the suggested system combines. The basis of the capability of motion detection in FMCW radars' is to recognize the Doppler shift in the received signal brought on by the target's motion. To properly identify the hand motions, the presented technique combines these two techniques. The system is analyzed using a collection of hand gesture photos, and the outcomes are analyzed with those of other hand gesture recognition systems which are already in use. A dataset of five different hand gestures is used to examine the proposed system. According to the experimental data, the suggested system can recognize gestures with an accuracy of 96.5%, showing its potential as a productive gesture recognition system. Additionally, the suggested system has a processing time of 100 ms and can run in real time. The outcomes also demonstrate the proposed system's resistance to noise and its ability to recognize gestures in a variety of configurations. For gesture detection applications in virtual reality and augmented reality systems, this research offers a promising approach.