Table 2 Cross-evaluation of mDice scores for various segmentation methods across five polyp segmentation datasets: Kvasir, CVC-ClinicDB, CVC-ColonDB, ETIS, and CVC-300. The model was trained on Kvasir-SEG and CVC-ClinicDB datasets with 1450 images, and tested on the others. The bolded values represent the highest mDice scores for each dataset

U-Net [8]

0.7615

0.8061

0.4734

0.5468

0.5969

U-Net++ [13]

0.8106

0.8769

0.4950

0.4762

0.6485

ResUNet++ [9]

0.7581

0.8603

0.4007

0.3701

0.5027

HarDNet [36]

0.8892

0.9054

0.6139

0.8228

0.8793

U²-Net [11]

0.8819

0.9204

0.6776

0.5543

0.7551

M²SNet [12]

0.9045

0.9059

0.6042

0.8111

0.8479

TransUNet [15]

0.9187

0.9259

0.7382

0.8246

0.8692

UTNet [37]

0.8844

0.9263

0.6341

0.5458

0.8076

PGCF [7]

0.9216

0.9362

0.7687

0.8249

0.8742

CoinNet [38]

0.8861

0.8852

0.6207

0.5782

0.7561

DCATNet

0.9266

0.9465

0.7872

0.8511

0.9064