Table 10. Success accuracy and loss values on the APTOS 2019 Blindness Detection Validation dataset.

Model	Metric	AF						AFCS-CNN
		ReLU	ELU	SELU	Mish	SiLU	GELU
VGG16	Accuracy (%)	76.91	69.09	75.27	80.73	77.82	77.09	80.18
	Loss	1.1004	0.8905	0.6830	0.6732	0.6521	1.0028	0.5188
VGG19	Accuracy (%)	78.73	74.91	75.45	76.18	73.82	78.18	78.73
	Loss	0.6350	0.7446	0.6775	0.6536	0.7559	0.5691	0.5652
DenseNet121	Accuracy (%)	81.82	74.00	76.36	77.45	72.91	78.55	83.27
	Loss	0.8153	0.7373	0.6211	0.6366	0.9755	0.7211	0.5611
DenseNet169	Accuracy (%)	79.09	64.36	57.64	69.45	77.09	80.73	83.45
	Loss	0.9562	1.0919	1.3097	0.9848	0.6766	0.7533	0.6115
EfficientNetV2B0	Accuracy (%)	62.55	74.18	59.27	74.91	70.91	70.91	79.45
	Loss	1.1221	0.9009	1.2356	0.9989	1.5562	1.0489	0.8403
EfficientNetV2B1	Accuracy (%)	69.27	62.55	66.91	79.64	73.27	73.45	81.82
	Loss	1.0717	1.2512	1.0693	1.1388	1.0198	1.0289	0.8520
ConvNeXtTiny	Accuracy (%)	No learning occurred in any of the trainings
	Loss
ConvNeXtSmall	Accuracy (%)	No learning occurred in any of the trainings
	Loss