Table 3 Comparison results on Kvasir-SEG.
From: CFM-UNet: coupling local and global feature extraction networks for medical image segmentation
Model | DICE | TPR | IoU | VOE | ASSD |
|---|---|---|---|---|---|
U-Net | 50.52% | 37.49% | 45.41% | 62.51% | 16.52% |
ATT-UNet | 50.75% | 37.35% | 45.99% | 62.65% | 16.13% |
TransUNet | 57.92% | 44.01% | 56.74% | 55.99% | 13.05% |
ResU-Net | 58.83% | 45.00% | 55.84% | 55.00% | 12.49% |
UltraLight VM-UNet | 62.59% | 50.32% | 68.04% | 49.68% | 10.33% |
VM-UNet-V2 | 66.95% | 54.70% | 77.40% | 45.30% | 7.72% |
M2SNet | 69.30% | 59.79% | 69.85% | 40.21% | 9.28% |
U-ResNet | 70.43% | 60.43% | 72.84% | 39.57% | 10.82% |
U-Net++ | 72.38% | 61.28% | 73.58% | 38.72% | 7.84% |
Swin-UMamba | 74.15% | 64.36% | 76.36% | 35.64% | 7.91% |
META-UNet | 76.26% | 67.86% | 79.16% | 32.14% | 7.16% |
CFM-UNet | 77.14% | 67.97% | 81.12% | 28.80% | 6.23% |
