Table 6.

CNN-based classifiers for comparison.

Classifier	Grading	Precision	Recall	F1-Score	Accuracy
ResNet50	I	0.786	0.786	0.786	0.778
	II	0.714	0.714	0.714
	III	0.833	0.833	0.833
MobileNetV2	I	0.727	0.571	0.640	0.690
	II	0.608	0.738	0.667
	III	0.762	0.762	0.762
Xception	I	0.676	0.595	0.632	0.619
	II	0.511	0.571	0.539
	III	0.690	0.690	0.690
DenseNet121	I	0.789	0.714	0.750	0.722
	II	0.644	0.690	0.667
	III	0.744	0.762	0.753
KNN	I	0.865	0.762	0.810	0.746
	II	0.675	0.643	0.659
	III	0.714	0.833	0.769
RF	I	0.791	0.810	0.800	0.786
	II	0.757	0.667	0.709
	III	0.804	0.881	0.841
NB	I	0.703	0.619	0.658	0.643
	II	0.583	0.500	0.583
	III	0.642	0.810	0.716
SVM	I	0.805	0.786	0.795	0.769
	II	0.689	0.738	0.713
	III	0.825	0.786	0.805
HCCANet	I	0.902	0.881	0.892	0.873
	II	0.850	0.810	0.829
	III	0.867	0.929	0.897