Table 1.
Comparison of the models’ sizes and the sizes of the data splits used for training, validation and testing
| Model | # Total | # Trainable | Training | Validation | Test | Ratio |
|---|---|---|---|---|---|---|
| parameters | parameters | Test to Train | ||||
| Mask R-CNN (segmentation) | 31.78M | 31.78M | 650 | 100 | 0.153 | |
| COVID-CT-Mask-Net (only S) | 31.52M | 2.25M | ||||
| COVID-CT-Mask-Net (S+BN) | 2.36M | 3K | 20.6K | 21.1K | 7.060 | |
| COVID-CT-Mask-Net (full) | 31.52M | |||||
| LightCNN [4] | 1.20M | 1.20M | 1528/1768 | 118/138 | 392/203 | 0.258/0.117 |
| COVNet [2] | 25.61M | 25.61M | 3K | 370 | 438 | 0.129 |
| ResNet18 [3] | 11.69M | 11.69M | 528 | 90 | 0.170 | |