inner workings of Mobile-net resolution multiplier - what does it do?

Question

i have a question concerning the way Mobile Net's resolution parameter works. From the article itself and from the blog posts on the topic (1, 2) I wasn't able to find an answer to my question. It is said that “the resolution multiplier changes the input dimensions of the image, reducing the internal representation at every layer”. Does that mean it (the parameter):

A) breaks down each image into a, for example, respective 224x224 bits of pixels sub-images, over which the network runs its convolution?

OR

B) takes one single 224x224 slice of pixels from, for example, the middle of the image, over which the network runs its convolutions?

OR

C) Squishes the original image into a, for example 224x224 pixel representation of itself, over which the network runs its convolutions?