Digital Forensics Use Case for Glaucoma Detection Using Transfer Learning Based on Deep Convolutional Neural Networks

Table 4

A comparison among different deep learning neural networks in terms of precision, recall, F1-score, and support.

CNN network	Class	Precision	Recall	F1-score	Support
DenseNet201	Healthy	0.50	0.00	0.00	1184.00
	Unhealthy	0.30	1	0.47	518
EfficientNetB7	Healthy	0.7	1	0.82	1184
	Unhealthy	0	0	0	518
InceptionV3	Healthy	0.7	0.97	0.81	1184
	Unhealthy	0.42	0.05	0.09	518
NASNetMobile	Healthy	0.61	0.23	0.33	1184
	Unhealthy	0.27	0.66	0.38	518
ResNet50	Healthy	0.7	0.57	0.63	1184
	Unhealthy	0.31	0.45	0.37	518
ResNet101V2	Healthy	0.64	0.07	0.13	1184
	Unhealthy	0.3	0.91	0.45	518
ResNet152	Healthy	0.7	1	0.82	1184
	Unhealthy	0.33	0	0.01	518
VGG16	Healthy	0.67	0.25	0.37	1184
	Unhealthy	0.29	0.71	0.42	518