Continuous Chinese sign language recognition and classification based on millimeter wave radar

Continuous sign language recognition has received widespread attention in the field of human-computer interaction. Compared with optical camera based continuous sign language recognition systems, millimeter wave radar based sign language recognition systems have unique advantages such as high integration, all-weather operation, and robustness in low light environments. This study constructed a continuous Chinese sign language recognition method using 60GHz Doppler radar, which utilizes a convolutional neural network incorporating Efficient Channel Attention (ECA) to process radar micro Doppler signals. A sign language radar data acquisition device was established on the Doppler radar, and 20 commonly used sign language statements for train stations were collected in the time-frequency domain to form a dataset. Subsequently, these data were input into the proposed neural network model for recognition and classification. The experiment shows that the proposed system achieves an accuracy of 97.92% in recognizing sign language at train stations. Compared with other radar based sign language recognition systems, the proposed system exhibits higher recognition accuracy.