Continuous Human Motion Recognition Based on FMCW Radar and Transformer

Figure 10

A typical set of data annotations. A set of data consisting of three actions is 12 seconds long. e divided the 12 seconds of data into 40 sequences, each representing 0.3 seconds, and contained only a single sequence within this sequence.