Application of U-Net with Global Convolution Network Module in Computer-Aided Tongue Diagnosis

Table 2

Model comparison of different convolutions in GCN module.

Model	PA (pixel accuracy)	Loss (%)	MIoU (mean intersection over union)
SEResNet_GCN_Unet	98.5	5.10	43.1
SEResNet_GCND_Unet	98.5	5.19	39.6
VGGNet_GCN_Unet	98.6	4.84	46.2
VGGNet_GCND_Unet	98.6	4.46	46.2
GoogLeNet_GCN_Unet	98.7	4.48	47.3
GoogLeNet_GCND_Unet	98.6	4.02	47.5