Data Augmentation-Assisted Makeup-Invariant Face Recognition

Table 1

Comparison of the proposed techniques.

Experiment No.	Augmentation	Dataset	Test accuracy
1	6 celebrity-famous makeup styles + 3 semantic preserving transformation (the proposed augmentation-assisted approach)	YMU	90.04
		VMU	92.99
2	No data augmentation is used	YMU	56.00
		VMU	57.85
3	6 celebrity-famous makeup styles	YMU	75.61
		VMU	78.19
4	3 semantic-preserving transformations	YMU	67.66
		VMU	68.05
5	Transfer learning without data augmentation	YMU	86.37
		VMU	88.48