Table 4.

Performance comparison of models trained with 0.001 learning rate and by using focal loss.

Model	Prefilter applied	Prefilter not applied
EfficientNet	%83.17	%83.76
EfficientNetV2	%81.34	%80.57
MobileNetV3Large	%78.69	%77.75
MobileNetV3Small	%81.39	%78.85
Resnet18	%81.47	%81.95
ShuffleNet	%75.21	%70.44