An Approach for Multi-Target Classification for FMCW 4D Millimeter-Wave Radar Imaging

In this paper, we present an approach to improve the effectiveness of multi-target classification for FMCW 4D millimeter-wave (mmW) radar point clouds. Our method involves initially grouping the point cloud using the DBSCAN algorithm, followed by segmentation into individual targets. Subsequently, these segmented targets are input into the modified PointNet++ architecture for classification. The results show that our method outperforms existing methods in the task of multi-target classification in 4D mmW radar.