Segmentation of Breast Tubules in H&E Images Based on a DKS-DoubleU-Net Model

Table 3

Comparison of the computational complexity of the models.

Model	Trainable params	FLOPs
DKS-DoubleU-Net	127.38 M	254.71 M
DoubleU-Net+DenseNet	127.24 M	254.42 M
DoubleU-Net+KSM	111.88 M	223.69 M
DoubleU-Net	111.73 M	223.40 M
U-Net	98.68 M	197.31 M
ResNet++	15.50 M	30.97 M
DeepLabV3+	155.88 M	311.76 M
HRNet	108.93 M	217.85 M