some Improved Models base on SINet, as my final design in college
Model Smeasure meanEm wFmeasure MAE
SINet_Official 0.758 0.796 0.517 0.054
SINet_Train0 0.766 0.790 0.522 0.055
SINet_Train1 0.765 0.822 0.586 0.048
SINetS 0.768 0.823 0.599 0.045
SINetS_RFDCT_Dec1 0.775 0.836 0.611 0.044
SINetS_RFDCTHalf2_Dec1 0.775 0.833 0.606 0.045
SINetS_RFFFTHalf2_Dec1 0.774 0.833 0.612 0.044
SINetS_RFDCTHalf_Dec3 0.774 0.837 0.613 0.043
SINetS_RF2Half_Dec3 0.775 0.837 0.619 0.043
SINetS_RF2FFTHalf_Dec3 0.775 0.840 0.614 0.043
SINetS_RF2FFTHalf2_Dec3 0.774 0.834 0.609 0.045
Model Loss Epoch DataAug Val Smeasure meanEm wFmeasure MAE
SINetv2(Official) SLoss 100 TRUE TRUE 0.815 0.887 0.680 0.037
SINetv2 BCE 64 FALSE FALSE 0.794 0.840 0.624 0.041
SI_Decs BCE 64 FALSE FALSE 0.797 0.847 0.639 0.040
SINetv2 SLoss 64 FALSE FALSE 0.790 0.879 0.650 0.040
SI_Decs SLoss 64 FALSE FALSE 0.805 0.886 0.679 0.037
SINetv2 BCE 100 FALSE FALSE 0.790 0.842 0.626 0.041
SI_Decs BCE 100 FALSE FALSE 0.799 0.856 0.643 0.039
SINetv2 SLoss 100 FALSE FALSE 0.788 0.875 0.644 0.041
SI_Decs SLoss 100 FALSE FALSE 0.800 0.884 0.675 0.037
SINetv2 BCE 100 TRUE FALSE 0.813 0.836 0.626 0.040
SI_Decs BCE 100 TRUE FALSE 0.822 0.856 0.662 0.036
SINetv2 SLoss 100 TRUE FALSE 0.810 0.882 0.672 0.038
SI_Decs SLoss 100 TRUE FALSE 0.822 0.893 0.701 0.034
SI_MinDecs SLoss 100 TRUE FALSE 0.822 0.896 0.702 0.034
see https://github.com/DengPingFan/SINet and https://github.com/GewelsJI/SINet-V2