A Facial Expression Recognition Method Using Improved Capsule Network Model

Table 1

Comparison with existing methods on the CK+ and multi-PIE datasets before and after illumination and posture preprocessing (%).

Methods	Preprocessing	CK+	Multi-PIE
Reference [23] algorithm	No	55.33	44.53
	Illumination	57.26	49.15
	Illumination and posture	59.16	52.62
Reference [15] algorithm	No	54.41	45.43
	Illumination	57.28	48.29
	Illumination and posture	61.05	55.17
Reference [19] algorithm	No	59.67	56.24
	Illumination	62.38	58.31
	Illumination and posture	63.51	60.68
Proposed	No	63.47	59.63
	Illumination	65.28	62.45
	Illumination and posture	67.31	65.31