Open-set Occluded Person Identification with mmWave Radar

We introduce a new multi-modal dataset comprising point cloud data from mmWave radar, RGB and depth images, collected from 23 human subjects. We use the TI IWR6843ISK-ODS radar for transmitting and receiving radar signals, paired with an Intel RealSense D435 camera for capturing RGB and depth images at the resolution of 420 × 240.  Considering the diverse electromagnetic wave absorption and reflection properties of different materials, our experiment employs three types of obstacles: a clothing rack, a poster, and a potted plant. To broaden the scope of scenarios and comprehensively evaluate radar sensor performance, walking data is collected using the same experimental configuration without obstacles.  To produce radar point clouds, we employ the Cell-Averaging Constant False-Alarm Rate (CA-CFAR) algorithm along with the maximum-energy ridge extraction method.  In our experiment, we have 4 scenarios, where each represents distinct occlusion conditions. Despite minimal data loss, thedataset contains over 300,000 frames of radar signals and over 600,000 RGB and depth images in total.