Table 5.

Accuracy (%) and F1 Score comparison of the SVM with other classifiers for individual RSI item classification on Caltech ADOS video datasets using “Top3” scenario fusion.

Classifier	B01		B02		B03		B04		B05
	Accu.	F1	Accu.	F1	Accu.	F1	Accu.	F1	Accu.	F1
MLP (Werbos, 1974)	78.33	0.75	40.56	0.39	78.06	0.75	59.44	0.55	61.67	0.61
ResNet50 (He et al., 2016)	78.61	0.75	47.78	0.44	78.89	0.75	61.67	0.54	57.50	0.56
SVM (Cortes, 1995)	83.06	0.78	45.00	0.45	83.06	0.78	68.89	0.65	52.50	0.53
Classifier	B06		B07		B08		B09		B10
	Accu.	F1	Accu.	F1	Accu.	F1	Accu.	F1	Accu.	F1
MLP (Werbos, 1974)	37.78	0.38	64.17	0.65	50.28	0.49	61.94	0.61	56.94	0.57
ResNet50 (He et al., 2016)	57.22	0.53	59.44	0.58	47.78	0.40	71.11	0.64	42.22	0.40
SVM (Cortes, 1995)	42.78	0.41	66.39	0.65	49.44	0.47	64.44	0.61	51.94	0.51
Classifier		B11			B12			B13
		Accu.		F1	Accu.		F1	Accu.		F1
MLP (Werbos, 1974)		69.17		0.65	50.00		0.50	52.22		0.52
RestNet50(He et al., 2016)		64.44		0.58	47.22		0.44	54.44		0.55
SVM (Cortes, 1995)		66.94		0.63	47.50		0.48	56.67		0.57

The best results are highlighted by bold.