Image: RF-Pose, MIT
MIT have progressed on previous research that used radio signals to detect human movement through walls by adding a neural network that accurately generates a skeleton representation of the human in real time. It can also identify the human with 83% accuracy.
Source: