Table 2.

Results for the CVC-ClinicDB dataset. Boldface numbers denote the highest values for each metrics.

Model	mDSC	mIoU	Recall	Precision
FCN–8S [21]	0.810	–	0.748	0.883
Multi-scale patch-based CNN [71]	0.813	–	0.786	0.809
FCN [72]	0.830	–	0.773	0.900
MultiResUNet [59]	–	0.821	–	–
cGAN [73]	0.872	0.795	–	–
U-Net [23]	0.878	0.788	0.787	0.933
Multiple encoder-decoder network [74]	0.889	0.894	–	–
PraNet [30]	0.898	0.840	–	–
PolypSegNet [28]	0.915	0.862	0.911	0.962
ResUNet++ with CRF [26]	0.920	0.890	0.939	0.846
Double U-Net [27]	0.924	0.861	0.846	0.959
FANet [34]	0.934	0.894	0.934	0.940
Focus U-Net	0.941	0.893	0.956	0.930