Infrared-based facial points tracking and action units detection in context of car driving simulator

EURASIP Journal on Image and Video Processing

Table 4 Precision and tracking fragmentation of the different facial points tracking methods for the MMI dataset

Rad	MC	TM	C ₁	C ₂	C ₃	C ₄	C ₅	C ₆	C ₇	C ₈	C ₉	C ₁₀	C ₁₁	C ₁₂	C ₁₃	C ₁₄	C ₁₅	C ₁₆	C ₁₇	C ₁₈	$\overset{Â¯}{MC}$
10%	P%	PF	57	59	65	63	64	61	58	62	53	54	57	54	58	66	69	71	70	68	61
		AAM	59	60	80	79	77	62	63	60	71	75	72	74	70	69	68	75	77	74	70
		HCPF-AAM	81	82	92	93	91	83	80	90	89	91	87	88	91	93	95	96	95	93	89
20%	P%	PF	52	50	54	53	56	49	51	55	56	54	52	53	50	59	57	57	58	60	54
		AAM	55	57	73	75	74	59	56	58	67	68	66	69	63	62	65	70	69	71	65
		HCPF-AAM	77	76	84	88	89	81	78	88	85	84	83	85	88	90	89	92	91	89	85
50%	P%	PF	47	48	51	50	52	49	50	51	54	52	47	49	48	52	54	51	50	57	50
		AAM	52	50	68	69	71	53	51	55	64	62	63	66	59	58	60	65	64	66	60
		HCPF-AAM	75	74	79	80	78	79	76	84	83	81	80	82	84	89	88	90	87	86	81
–	TF%	PF	24	23	21	19	19	20	22	25	25	20	21	22	23	19	20	16	18	19	20
		AAM	23	24	19	18	18	22	21	19	19	20	21	19	18	19	18	16	19	17	19
		HCPF-AAM	12	10	7	7	8	11	10	7	7	8	9	10	10	7	5	5	4	5	8

Rad, radius circle size; TM, tracking method; MC, metric; $\overset{Â¯}{MC}$ , average of metric values for all facial points; C₁ to C₁₈, facial points shown in Figure 2.