Deep Neural Network for Accurate Age Group Prediction through Pupil Using the Optimized UNet Model

Table 12

Hyperparameters of the proposed system.

Dataset	Age grp	Size of the data	Rate of accuracy	Overall classification rate
Test data
	‘10–20’	60	80%	88.6%
	‘21–27’	50	86%
	‘28–45’	100	98%
	‘46–65’	50	82%
	‘66–100’	50	97%
Training data
CASIA	‘10–20’	20	94.7	89.6
	‘21–27’	20	88.5
	‘28–45’	24	89.6
	‘46–65’	36	84.5
	‘66–100’	14	83.6
UBER IS	‘10–20’	30	89.4	95.3
	‘21–27’	25	98.3
	‘28–45’	21	97.3
	‘46–65’	35	96.8
	‘66–100’	29	94.9
MMU	‘10–20’	10	87.6	93.1
	‘21–27’	10	89.3
	‘28–45’	20	97.4
	‘46–65’	20	95.4
	‘66–100’	25	96.1
Random dataset	‘10–20’	60	97.5	95.62
	‘21–27’	50	89.3
	‘28–45’	100	95.3
	‘46–65’	50	98.4
	‘66–100’	60	97.6
Avg				93.40