Evaluation of YOLOv3-based Radar Data Processing for Indoor Positioning: Application of mmDetect

In an indoor environment, determining the presence of people, their location and tracking their movements is a current research challenge. In this application field, Radars are devices that can be used for detection, localization and tracking. The main purpose of this paper is to test a method that takes advantage of all the information that can be obtained from a Frequency Modulated Continuous Wave (FMCW) Radar, through the application of the You Only Look Once (YOLO) network used to detect, locate and track the target inside the environment. The goal is to test the performances of the proposed algorithm and to verify the result provided by the network in determining the presence of a subject and its position in a room. The criticalities in the setup such as the positioning of the Radar and maintaining its stability during the tests will be discussed and also, a performance analysis will be provided to understand the capabilities of the proposed approach.