Multi-target identity recognition and behavior monitoring method based on millimeter wave radar

A multi-target identity recognition and behavior monitoring method based on a millimeter-wave radar comprises the following steps: step 1, constructing target gait identity features, realizing identity recognition based on a deep convolutional network, collecting millimeter-wave radar data, constructing a two-dimensional gait feature image through a series of signal processing technologies, and inputting the two-dimensional gait feature image into an OfficientNet-B0 network for identity recognition; 2, extracting point cloud data, improving a PointNet network to construct an attitude estimation model, extracting time sequence attitude data, and inputting the time sequence attitude data into a time convolution network TCN to realize behavior recognition; and 3, constructing an identity tag for behavior monitoring based on the identity information in the step 1, and constructing an integrated identity recognition and behavior monitoring system. The attitude estimation network and the behavior detection network