Hello All,
I have a section in my network which needs constant size inputs and I have variable size data and therefore I am using adaptive avg pool2D
. Mostly the image sized would get downscaled but there might be certain cases where image sizes go up let us say from (120, 120)
to (160, 160)
. So how does the function takes care of such cases. I believe it would pad the image internally, so what is the kind of padding?. And Secondly, is upscaling like this advisable?
TIA