How resize and zoom pre- and post- transform would work to keep same dimension?

[OmarAshkar]

Hello,

I tried to run inference on WSI, but the process is extermely slow. Now I am thing to downsample my data. I am wondering if Resize then Zoom transform would keep the same pixel locations to the original place or not?

Thanks in advance!

[KumoLiu]

Hi @omashkartrx, what's your pixel locations mean? Both of these transforms change the resolution of the output image if you are referring to the resolution.
Additionally, they are both using torch.nn.interpolate, for more details you could also see https://pytorch.org/docs/stable/generated/torch.nn.functional.interpolate.html.

Hope it can help you, thanks!