Table 5.

Metrics of the network trained with different types of encoders.

Exp.	Encoder	Acc	F1	IoU	Precision	Recall
1	EFFNet-b0	0.9942	0.9540	0.9113	0.9171	0.9942
2	EFFNet-b1	0.9179	0.9417	0.8901	0.9669	0.91795
3	EFFNet-b2	0.9129	0.9369	0.8817	0.9624	0.9129
4	EFFNet-b3	0.9115	0.9441	0.8944	0.9792	0.9115
5	EFFNet-b4	0.9350	0.9460	0.8979	0.9573	0.9350
6	EFFNet-b5	0.9584	0.9406	0.8883	0.9236	0.9584
7	EFFNet-b6	0.9302	0.9537	0.9117	0.9785	0.9302
8	EFFNet-b7	0.9285	0.9519	0.9085	0.9767	0.9285
9	MVT-0	0.8979	0.9365	0.8808	0.9787	0.8979
10	MVT-1	0.8717	0.9247	0.8603	0.9849	0.8717
11	MVT-2	0.9264	0.9477	0.9007	0.9700	0.9264
12	MVT-3	0.9407	0.9588	0.9210	0.9777	0.9407
13	MVT-4	0.9307	0.9551	0.9141	0.9808	0.9307
14	MVT-5	0.8800	0.9291	0.8681	0.9844	0.8800
15	ResNet34	0.9572	0.9587	0.9210	0.9603	0.9572
16	ResNet50	0.9436	0.9591	0.9206	0.9752	0.9436
17	ResNet101	0.9160	0.9461	0.8980	0.9784	0.9160
18	ResNet152	0.9217	0.9494	0.9039	0.9789	0.9217

The abbreviation MVT refers to Mix-vision-transform. EFFNet refers to EfficientNet.