Classification of Human Posture from Radar Returns Using Ultra-Wideband Radar

Distance and Angle Insensitive Radar-Based Multi-Human Posture Recognition Using Deep Learning

Human posture recognition has a wide range of applicability in the detective and preventive healthcare industry. Recognizing posture through frequency-modulated continuous wave (FMCW) radar poses a significant challenge as the human subject is static. Unlike existing radar-based studies, this study proposes a novel framework to extract the postures of two humans in close proximity using FMCW radar point cloud. With radar extracted range, velocity, and angle information, point clouds in the Cartesian domain are retrieved. Afterwards, unsupervised clustering is implemented to segregate the two humans, and finally a deep learning model named DenseNet is applied to classify the postures of both human subjects. Using four base postures (namely, standing, sitting on chair, sitting on floor, and lying down), ten posture combinations for two human scenarios are classified with an average accuracy of 96%. Additionally, using the centroid information of human clusters, an approach to detect and classify overlapping human participants is also introduced. Experiments with five posture combinations of two overlapping humans yielded an accuracy of above 96%. The proposed framework has the potential to offer a privacy-preserving preventive healthcare sensing platform for an elderly couple living alone.

A Real-Time Respiration Monitoring and Classification System Using a Depth Camera and Radars

Respiration rate (RR) and respiration patterns (RP) are considered early indicators of physiological conditions and cardiorespiratory diseases. In this study, we addressed the problem of contactless estimation of RR and classification of RP of one person or two persons in a confined space under realistic conditions. We used three impulse radio ultrawideband (IR-UWB) radars and a 3D depth camera (Kinect) to avoid any blind spot in the room and to ensure that at least one of the radars covers the monitored subjects. This article proposes a subject localization and radar selection algorithm using a Kinect camera to allow the measurement of the respiration of multiple people placed at random locations. Several different experiments were conducted to verify the algorithms proposed in this work. The mean absolute error (MAE) between the estimated RR and reference RR of one-subject and two-subjects RR estimation are 0.61±0.53 breaths/min and 0.68±0.24 breaths/min, respectively. A respiratory pattern classification algorithm combining feature-based random forest classifier and pattern discrimination algorithm was developed to classify different respiration patterns including eupnea, Cheyne-Stokes respiration, Kussmaul respiration and apnea. The overall classification accuracy of 90% was achieved on a test dataset. Finally, a real-time system showing RR and RP classification on a graphical user interface (GUI) was implemented for monitoring two subjects.