Fall detection based on millimeter-wave radar point cloud

Abstract With the intensification of population aging, the health issues of middle-aged and elderly people are becoming increasingly prominent, among which falls pose a particularly serious threat to their health. Therefore, to accurately detect falling actions, this paper proposes a fall detection method based on millimeter-wave radar. This method utilizes millimeter-wave radar to collect three-dimensional (3D) point cloud information and constructs a detection model based on the convolutional neural networks (CNN)-Long short-term memory network (LSTM)-Attention network. The fall detection approach utilizing the CNN-LSTM-Attention network primarily involves two key stages: the first step involves extracting the point cloud data of human activities from radar data through 3D fast Fourier transform (FFT), clutter filtering, and clustering algorithms. The second step is to train the CNN-LSTM-Attention model using CNN, LSTM, and attention mechanisms to extract the features of human activities more deeply and classify them. The experimental results indicate that the model’s accuracy in detecting human falls can reach 95.56%, which can effectively distinguish different human actions and verify the feasibility of detecting human falls using 3D point cloud information collected by millimeter-wave radar.