Difference between --img-size and --input-size

[AlejandroRigau]

I have my own dataset with images size 10x10 and I want to train Resnet and EfficientnetV2 models. What would be the difference between --img-size and --input-size? Do I need to set both to:
--img-size 10
--input-size 3 10 10 \

Is this the right way to do it? Can this be dont with --pretrained as well?