Regarding RGB vs BGR for vgg model

[InnovArul]

Hi,

In your code, you seem to use the image as BGR, whereas in the torchvision documentation, itseems VGG model expects RGB.
I have used your code and tried to use the image as RGB. But the model prediction is not confident (<0.5).

Do you know which one of the formats we have to use for PyTorch models(BGR or RGB)?