YOLO-UNet Architecture for Detecting and Segmenting the Localized MRI Brain Tumor Image

Table 5

The comparison of loss and accuracy for each scenario.

	YOLOv3-UNet				YOLOv4-UNet
	Training		Validation		Training		Validation
	Loss	Accuracy	Loss	Accuracy	Loss	Accuracy	Loss	Accuracy
Model 1	0.0103 (2)	0.9885 (4)	0.0126 (2)	0.9894 (3)	0.0553 (2)	0.9743 (3)	0.1211 (2)	0.9776 (1)
Model 2	0.0096 (1)	0.9899 (1)	0.0135 (3)	0.9895 (2)	0.0587 (4)	0.9740 (4)	0.1059 (1)	0.9770 (2)
Model 3	0.0111 (3)	0.9893 (2)	0.0124 (1)	0.9896 (1)	0.0576 (3)	0.9762 (2)	0.1706 (3)	0.9776 (1)
Model 4	0.0120 (4)	0.9887 (3)	0.0140 (4)	0.9891 (4)	0.0541 (1)	0.9777 (1)	0.1754 (4)	0.9776 (1)

(1/2/3/4) indicates the rank of lost and accuracy; (1) is the first rank and so on. Bold value indicates the optimum number of loss and accuracy.