Table 2 The test results of the compare study in Kvasir and CVC-ClinicDB.

From: Dual ensemble system for polyp segmentation with submodels adaptive selection ensemble

Train/Val

Kvasir 80%/10%

CVC-ClinicDB 80%/10%

Test

Kvasir 10%

CVC-ClinicDB 10%

Metric

Dice

mDice

mIoU

p

Dice

mDice

mIoU

p

UNet

0.814

0.857

0.822

< 0.05

0.871

0.87

0.876

< 0.05

UperNet

0.855

0.884

0.861

< 0.05

0.882

0.913

0.886

< 0.05

Double-UNet

0.794

0.824

0.797

< 0.05

0.832

0.841

0.841

< 0.05

SS-Former

0.898

0.940

0.896

< 0.05

0.923

0.958

0.929

< 0.05

FCB-Former

0.901

0.942

0.904

0.15

0.944

0.969

0.945

0.26

ESFPNet

0.896

0.938

0.898

< 0.05

0.945

0.971

0.946

0.41

Nanni’s Ens

0.916

0.948

0.912

0.35

0.946

0.973

0.949

0.69

HarDNet-DFUS

0.908

0.943

0.905

0.18

0.943

0.969

0.944

0.21

DET-Former

0.923

0.954

0.924

_

0.953

0.975

0.953

_

Back to article page