CNN-Based Automatic Helmet Violation Detection of Motorcyclists for an Intelligent Transportation System

Table 4

Experimental results.

Performance metrics	Scores	Derivations
Sensitivity	0.9825	TPR = TP/(TP + FN)
Specificity	0.9694	SPC = TN/(FP + TN)
Precision	0.9770	PPV = TP/(TP + FP)
Negative predictive value	0.9767	NPV = TN/(TN + FN)
False positive rate	0.0306	FPR = FP/(FP + TN)
False discovery rate	0.0230	FDR = FP/(FP + TP)
False negative rate	0.0175	FNR = FN/(FN + TP)
Accuracy	0.9769	ACC = (TP + TN)/(P + N)
F1 score	0.9798	F1 = 2TP/(2TP + FP + FN)
Matthews correlation coefficient	0.9528	TP × TN − FP × FN/sqrt((TP + FP) × (TP + FN) × (TN + FP) × (TN + FN))