An Efficient Approach of Face Detection and Prediction of Drowsiness Using SVM

Table 5

Performance of the models VGG16, VGG19, RESNET50, RESNET101 and proposed model including different dataset.

S.No.	Technique	Dataset	Epochs	Accuracy
1	VGG 16	ZJU [22]	100	98.15
		MRL [23]	100	92.56
		Proposed dataset	100	98.68
2	VGG19	ZJU [22]	100	94.96
		MRL [23]	100	90.23
		Proposed dataset	100	98.74
3	RESNET50	ZJU [22]	100	62.50
		MRL [23]	100	60.33
		Proposed dataset	100	95.69
4	RESNET101	ZJU [22]	100	65.89
		MRL [23]	100	62.31
		Proposed dataset	100	95.77
5	MobileNetV2	ZJU [22]	100	84.70
		MRL [23]	100	95.88
		Proposed dataset	100	96.00
6	Proposed model with SVM	ZJU [22]	100	97.54
		MRL [23]	100	98.76
		Proposed dataset	100	99.56
			500	99.85