On Study of 1D Depth Scans as an Alternative Feature for Human Pose Detection in a Sensor Network

Table 3

The accuracy of orientation detection in different scenarios where the training set is capture data 1 m from the sensor plane and the test is from a different subject in distance 1 m in scenario 1, and the captured test set is in a distance of 1 m, 2 m, and 3 m from sensors’ plane in Sc2, Sc3, and Sc4, respectively.

Number of FFT profile	Model	Sc 1	Sc 2	Sc 3	Sc 4
	KNN	51.13	62.61	52.94	63.24
	CNN	64.31	71.30	62.75	62.50
	KNN	53.38	61.74	64.71	83.09
	CNN	6817	80.87	79.74	58.09
	KNN	7138	95.65	90.20	46.32
	CNN	72.99	77.39	66.67	38.24
	KNN	80.39	94.78	70.59	38.24
	CNN	7846	84.35	60.78	38.24
	KNN	85.53	92.17	56.86	38.24
	CNN	81.35	85.22	56.21	38.24
	KNN	86.82	88.70	56.21	38.24
	CNN	82.32	84.35	52.94	38.24
	KNN	85.85	86.09	56.86	38.24
	CNN	83.92	82.61	49.67	38.24
	KNN	85.53	8783	56.86	38.24
	CNN	83.60	74.78	49.02	38.24
	KNN	84.24	87.83	56.86	38.24
	CNN	81.99	73.91	49.67	38.24
	KNN	83.28	88.70	58.17	38.24
	CNN	81.99	76.52	52.29	38.24
	KNN	83.92	90.43	56.86	38.24
	CNN	81.99	79.13	52.94	38.24
	KNN	83.92	89.57	58.17	38.24
	CNN	81.99	81.74	53.59	38.24