TABLE 2.

Summary of an evaluation matrix for semantic segmentation of U-Net using various deep learning backbones (encoder).

Backbones	Model evaluation matrix		Validation matrix
	IoU	F1-score	IoU	F1-score
VGG16	0.9384	0.9682	0.9272	0.9626
VGG19	0.9464	0.9724	0.9301	0.9637
SEResNet152	0.9665	0.9824	0.9323	0.9639
SEResNeXt101	0.9684	0.9839	0.9324	0.9648
SENet154	0.9697	0.9846	0.9314	0.9643
ResNet154	0.9565	0.9777	0.9259	0.9613
ResNeXt101	0.9623	0.9808	0.9281	0.9614
MoblieNetV2	0.9518	0.9749	0.9250	0.9608
InceptionResNetV2	0.9640	0.9817	0.9308	0.9637
DenseNet201	0.9609	0.9801	0.9310	0.9642

Two evaluation scores are shown in the table. Intersection-over-union (IoU) evaluation matrix and F1-score were calculated.