Table 1.

Characteristics of the networks used in the evaluation.

Network	Characteristics
ResNet50 [22]	pixel size $\left(250\times 250\times 3\right)$, batch size = $64$, dropout = $0.2$, global average pooling, dense nodes $1024 \times 1024$, epochs = $200$
VGG16 [35]	pixel size $\left(250\times 250\times 3\right)$, batch size = $32$, dropout = $0.2$, flatten, dense nodes $512 \times 512$, epochs = $200$
MobileNet [39]	pixel size $\left(300\times 300\times 3\right)$, batch size = $32$, dropout = $0.5$, global average pooling, epochs = $200$
InceptionV3 [36]	pixel size $\left(250\times 250\times 3\right)$, batch size = $32$, dropout = $0.7$, global average pooling, dense nodes = $1500\times 1500$, epochs = $200$
Xception [37]	pixel size $\left(300\times 300\times 3\right)$, batch size = $16$, dropout = $0.2$, flatten, dense nodes = $512\times 512$, epochs = $200$
Papadrianos et al. [14]	pixel size $\left(400\times 400\times 3\right)$, batch size = 32, dropout = 0.2, $16–32–64–128$ dense nodes =$32, 16$, epochs = $300$
LB-FCN light [33]	pixel size $\left(224\times 224\times 3\right)$, batch size = 32, global average pooling, epochs = $200$