Table 5.
Comparison with state-of-the-art methods on multi-class classification.
| Method | Data distribution | Accuracy % |
|---|---|---|
| Transfer Learning (on Dataset 1 in cited paper) (2020) [13] | 224 COVID-19, 700 Pneumonia, 504 Normal | 93.48 |
| Transfer Learning (on Dataset 2 in cited paper) (2020) [13] | 224 COVID-19, 714 Pneumonia, 504 Normal | 94.72 |
| DarkCovidNet (2020) [29] | 127 COVID-19, 500 Pneumonia, 500 Normal | 87.02 |
| Majority Voting ML (2020) [24] | 782 COVID-19, 782 Pneumonia, 782 Normal | 93.41 |
| DenseNet-201 (2020) [14] | 423 COVID-19, 1485 Pneumonia, 1579 Normal | 97.94 |
| VGG16 (2020) [15] | 142 COVID-19, 142 Pneumonia, 142 Normal | 95.88 |
| CovXNet (2020) [19] | 305 COVID-19, 2780 Bact. Pneumonia, 1493 Viral Pneumonia, 1583 Normal | 90.2 |
| CNN + SVM (2021) [36] | 77 COVID-19, 256 Normal | 99.02 |
| Cascaded CNNs (2020) [18] | 69 COVID-19, 79 Bact. Pneumonia, 79 Viral Pneumonia, 79 Normal | 99.9 |
| CoroNet (on Dataset 1 in cited paper) (2020) [17] | 284 COVID-19, 657 Pneumonia, 310 Normal | 95.0 |
| CoroNet (on Dataset 2 in cited paper) (2020) [17] | 157 COVID-19, 500 Pneumonia, 500 Normal | 90.21 |
| Pruned Weighted Average (2020) [22] | 313 COVID-19, 8792 Pneumonia, 7595 Normal | 99.01 |
| FFB3 (2021) [37] | 125 COVID-19, 500 pneumonia, 500 no-finding, and | 87.64 |
| Deep-CNN (2021) [38] | 2161 COVID-19, 2022 pneumonia, and 5863 normal chest | 92.63 |
| ResNet-50 and AlexNet (2021) [63] | 3,616 COVID-19, 1,345 pneumonia, 10,192 normal, and 6,012 lung opacity | 95 |
| Blended ensembling on COVID-X | 568 COVID-19, 6052 Pneumonia, 8851 Normal | 94.55 |
| Blended ensembling on Chowdhury et al. dataset[14] | 219 COVID-19, 1345 Pneumonia, 1341 Normal | 98.13 |