Temporal Convolutional Network with Wavelet Transform for Fall Detection

Table 1

Overall feature of three datasets.

Datasets	ADLs	Falls	Data source	Sampling rates	Location	Subject numbers	Age	Gender
								Female	Male
UniMiB SHAR	Yes	Yes	A Samsung phone with a BMA220 acceleration sensor	50 Hz	Trouser pockets	30	18 - 60	24	6
SisFall	Yes	Yes	A self-developed embedded device composed of a CPU, a ADXL345 accelerometer	200 Hz	Waist	38	19 - 75	19	19
UMAFall	Yes	Yes	An android smartphone and a set of mobility sensors attached to different parts of the body	200 Hz	Chest, waist, wrist, ankle	17	14 - 55	6	11