Radar as a Security Measure - Real Time Neural Model based Human Detection and Behaviour Classification

Abstract

Radar technology has great potential for use as security systems in college campuses with privacy advantages over its visual counterparts. We experimented with the Texas Instruments IWR1642 radar sensor to evaluate the feasibility of using radar systems for security monitoring. We introduce an end-to-end neural architecture which is capable of taking radar data inputs in real time and identify human versus nonhuman targets, and classify various human behavioral motions. The model is real time, robust to noise and displays state-of-the-art results for the problem.