Table 5 Ablation Study for Classification (ISIC 2020 Dataset).

From: HyperFusionNet combines vision transformer for early melanoma detection and precise lesion segmentation

Method

Params (M)

Accuracy

ROC AUC

HyperFusion-Net (Full)

3.15

0.9667

0.9701

HyperFusion-Net w/o Fusion Block

3.10

0.9333

0.9350

HyperFusion-Net w/o Attention Gates

3.12

0.9467

0.9480

HyperFusion-Net w/ Single-Path ViT

2.95

0.9400

0.9412

Back to article page