“…In addition, Law and Lin [44] implemented Transfer learning on a dataset with 1200 images of COVID-19 patients and found that VGG-16 gives superior performance metrics as compared to the other ResNet models. A Multi-model fusion study was performed by Cengil and Cinar [45] , tested various models namely AlexNet, EfficientNet-b0, NASNetLarge and Xception to find the best performing amalgamations. Khan et al [23] used pre-trained Xception architecture to get 95% 3-class and 89.6% 4-class accuracy on a dataset of 284 COVID-19, 310 healthy, 227 Viral Pneumonia and 330 Bacterial Pneumonia samples.…”