A reliable diabetic retinopathy grading via transfer learning and ensemble learning with quadratic weighted kappa metric

Table 3 Metrics scores versus the model

Model	Accu.	F1-score	Prec.	QWK	Recall	AUC-ROC score
VGG-16	0.732	0.625	0.533	0	0.726	0.196
VGG-19	0.734	0.617	0.544	0	0.732	0.196
ResNet-18	0.723	0.614	0.525	0	0.714	0.196
ResNet-34	0.724	0.612	0.512	0	0.723	0.196
ResNet-50	0.735	0.623	0.534	0	0.726	0.196
ResNet-101	0.766	0.613	0.556	0.009	0.73	0.196
ResNet-152	0.721	0.612	0.578	0.067	0.723	0.196
EffNet-B0	0.767	0.721	0.689	0.501	0.767	0.332
EffNet-B1	0.778	0.722	0.689	0.478	0.778	0.312
EffNet-B2	0.767	0.712	0.667	0.502	0.767	0.267
EffNet-B3_30	0.852	0.844	0.833	0.824	0.845	0.478
EffNet-B3_60	0.867	0.842	0.854	0.854	0.867	0.601
EffNet-B3_En	0.924	0.867	0.867	0.867	0.878	0.742
EffNet-B4	0.767	0.701	0.694	0.401	0.767	0.332
EffNet-B5	0.721	0.612	0.521	0	0.722	0.343
EffNet-B6	0.767	0.723	0.701	0.532	0.767	0.312

ISSN: 1472-6947