Skip to main content
. 2022 Jun 15;14(12):2956. doi: 10.3390/cancers14122956

Table 4.

Performance comparisons of various deep automatic segmentation models.

Model Num_params Num_iters Total Training Time/s Convergence Time/s
MVI-Mind (our work) 84,596,418 100,000 63,075 about 6550
Swin transformer 108,235,650 100,000 78,420 about 15,680
DeepLab V3+ 45,871,090 100,000 40,218 about 3890
U-Net 13,404,354 100,000 18,930 about 1520