Table 6 Results of comparison experiments on the Chaoyang dataset.
From: Multi-branch CNN and grouping cascade attention for medical image classification
Method(year) | Params (M) | FLOPs (G) | Acc | F1 | Precision | Recall | Auc |
|---|---|---|---|---|---|---|---|
ResNet50 (2016) | 23.5 | 4.1 | 0.8266 | 0.7783 | 0.7740 | 0.7849 | 0.8643 |
MobileNetV2 (2018) | 2.2 | 0.3 | 0.8242 | 0.7702 | 0.7738 | 0.7680 | 0.8547 |
EfficientNet-B0 (2019) | 4.0 | 0.4 | 0.8532 | 0.8021 | 0.8074 | 0.7986 | 0.8745 |
RepVGG (2021) | 43.7 | 9.9 | 0.8532 | 0.7999 | 0.8043 | 0.7960 | 0.8736 |
ConvNext-S (2022) | 49.5 | 8.7 | 0.7835 | 0.7189 | 0.7155 | 0.7242 | 0.8262 |
ConvMixer (2023) | 47.9 | 49.1 | 0.8340 | 0.7736 | 0.7827 | 0.7691 | 0.8562 |
InceptionNext-S (2023) | 47.1 | 8.4 | 0.8481 | 0.7970 | 0.8023 | 0.7925 | 0.8705 |
FasterNet (2023) | 13.7 | 1.9 | 0.8439 | 0.7936 | 0.8025 | 0.7889 | 0.8682 |
Swin-S (2021) | 48.8 | 8.6 | 0.8513 | 0.8029 | 0.8109 | 0.7983 | 0.8744 |
CrossViT 18 (2021) | 43.3 | 9.0 | 0.8125 | 0.7529 | 0.7674 | 0.7436 | 0.8391 |
BiFormer-S (2023) | 56.0 | 9.4 | 0.8312 | 0.7628 | 0.7860 | 0.7591 | 0.8503 |
FasterViT-2 (2023) | 75.2 | 8.9 | 0.8320 | 0.7577 | 0.7745 | 0.7483 | 0.8454 |
Flatten-pvt (2023) | 24.2 | 3.7 | 0.8396 | 0.7877 | 0.8035 | 0.7766 | 0.8605 |
TransxNet (2023) | 25.5 | 4.6 | 0.8439 | 0.7886 | 0.8039 | 0.7786 | 0.8627 |
GroupMixFormer (2023) | 22.1 | 5.1 | 0.8537 | 0.7971 | 0.8167 | 0.7844 | 0.8667 |
Eff-CTNet(Ours) | 25.2 | 6.4 | 0.8635 | 0.8090 | 0.8191 | 0.8012 | 0.8776 |
