Temporal Convolutional Network with Wavelet Transform for Fall Detection

Figure 1

The workflow of WTCN. (a) The first step to input raw signals from sensor-based datasets. Then, through wavelet transform (step b), 1D signals could be converted to 2D images. (c) About model structuring and training procedures based on an optimized TCN network. Finally, we can detect the fall behaviors through the above steps.