Fall Detection Using Ultra-Wideband Radar and Capsule Network
We propose a fall detection method from radar time-frequency representations. A radar time series is derived from the ultra wideband radar return signals. The time-frequency representations are obtained by applying STFT. The resulting spectrograms are used as input to the deep capsule network for automatic feature learning.
H. Sadreazami, M. Bolic and S. Rajan, CapsFall: Fall Detection Using Ultra-Wideband Radar and Capsule Network, IEEE ACCESS, 2019.
If you find the script helpful, please cite the the paper.
Contact me if you have questions.