Table 4 The performance of emotion recognition was evaluated independently in each modality.

Sub.	Gender	EEG		Audio		Visual
Sub.	Gender	ACC [%]	F-score	ACC [%]	F-score	ACC [%]	F-score
1	M	67.5	0.66	49.2	0.48	37.5	0.38
2	F	80.0	0.78	73.3	0.73	55.8	0.57
3	F	52.5	0.52	67.5	0.67	76.7	0.76
4	M	77.5	0.77	71.7	0.71	79.2	0.79
5	M	52.5	0.52	65.8	0.66	74.2	0.73
6	F	47.5	0.47	77.5	0.78	74.2	0.73
7	F	59.2	0.59	67.5	0.67	77.5	0.76
8	F	56.7	0.55	55.8	0.56	75.8	0.75
9	M	44.2	0.44	70.0	0.69	74.2	0.74
10	M	66.7	0.66	55.0	0.53	64.2	0.63
11	M	46.7	0.46	66.7	0.66	63.3	0.60
12	M	65.8	0.67	52.5	0.53	52.5	0.53
13	F	64.2	0.63	57.5	0.58	80.0	0.80
14	M	47.5	0.48	66.7	0.67	55.0	0.54
15	F	59.2	0.58	66.7	0.65	83.3	0.82
16	F	55.0	0.50	60.8	0.61	61.7	0.55
17	M	78.3	0.78	80.0	0.80	96.7	0.97
18	M	69.2	0.68	66.7	0.64	75.0	0.74
19	F	55.0	0.54	63.3	0.63	77.5	0.77
20	F	77.5	0.77	66.7	0.67	79.2	0.79
21	F	68.3	0.66	59.2	0.57	75.0	0.75
22	F	76.7	0.77	68.3	0.66	76.7	0.76
23	M	63.3	0.62	68.3	0.69	61.7	0.59
24	F	75.0	0.70	50.8	0.51	73.3	0.72
25	F	49.2	0.50	66.7	0.66	70.0	0.69
26	F	58.3	0.56	59.2	0.59	81.7	0.80
27	F	70.0	0.70	64.2	0.63	88.3	0.88
28	F	63.3	0.63	59.2	0.59	83.3	0.83
29	M	51.7	0.49	40.8	0.39	78.3	0.77
30	F	59.2	0.58	61.7	0.63	68.3	0.67
31	M	48.3	0.48	58.3	0.58	67.5	0.62
32	F	60.0	0.60	51.7	0.52	62.5	0.59
33	F	73.3	0.72	84.2	0.84	87.5	0.88
34	F	54.2	0.48	44.2	0.42	77.5	0.76
35	F	30.8	0.29	48.3	0.46	63.3	0.63
36	M	58.3	0.54	65.0	0.62	65.0	0.63
37	F	54.2	0.50	45.0	0.45	60.0	0.60
38	F	64.2	0.62	52.5	0.51	82.5	0.82
39	F	59.2	0.58	68.3	0.69	64.2	0.64
40	M	40.0	0.39	61.7	0.60	58.3	0.52
41	F	26.7	0.23	65.0	0.64	85.8	0.86
42	F	72.5	0.73	50.8	0.54	79.2	0.80
Mean		60.0	0.58	61.9	0.61	71.4	0.70

The accuracy and f-scores for 5 balanced classes were calculated across individual subjects. The represented CNN methods were EEGNet, DeepFaceNet, and SCNN for EEG, Visual, and Audio data, respectively.

Quick links

Search