Research Article
Multiscale Feature Fusion for Skin Lesion Classification
Table 7
Comparison with other existing methods.
| | Method | Date | Precision (%) | Recall (%) | Para (M) | Accuracy (%) | Macroaverage (%) |
| | DenseNet, SENet, ResNeXt [19] | 2018 | — | — | — | 85.10 | — | | InceptionV3+ResNet-50 [20] | 2018 | 86.2 | 79.60 | — | 89.90 | — | | MSM-CNN [21] | 2020 | 91.30 | — | — | 86.20 | 98 | | ResNet-50 [22] | 2020 | — | 81.00 | 23.54 | 89.28 | — | | ResNeXt, SeResNeXt, ResNet, Xception, DenseNet [23] | 2021 | 87.00 | 94.00 | — | 88.00 | 93 | | RegNetY-3.2GF-Drop [38] | 2021 | — | — | 15.30 | 86.40 | 97 | | Our method | | 83.53 | 95.04 | 19.70 | 91.24 | 98 |
|
|
