TABLE 4.

Accuracy comparison against baselines. The proposed method achieves the highest accuracy on all sequences. Although the gait cycle method has an accuracy closer to the proposed method, it does not remove false positives when gait cycle is violated.

Sequence	Method	% Accuracy					% False Positive Rate	% False Negative Rate
		Leg RF	Leg LF	Leg RH	Leg LH	Leg Avg	Leg Avg	Leg Avg
Concrete Short Loop	GRF Thresholding	73.43	70.02	71.69	70.04	71.30	37.07	13.24
	Gait Cycle	85.66	84.98	84.68	85.11	85.11	22.95	0.00
	Proposed Method	98.34	97.87	97.95	98.56	98.18	1.45	2.51
Grass Test Sequence	GRF Thresholding	82.55	78.93	84.62	82.48	82.14	26.87	0.63
	Gait Cycle	92.41	92.38	91.04	90.55	91.59	10.95	3.53
	Proposed Method	98.08	97.57	97.73	97.73	97.78	2.35	1.98
Forest Test Sequence	GRF Thresholding	80.99	80.09	82.75	83.24	81.77	26.54	1.84
	Gait Cycle	83.03	82.56	84.44	84.28	83.58	24.71	0.08
	Proposed Method	97.05	96.62	97.24	97.40	97.08	2.82	3.12

Bold values in tables show the best performance.