Slide to add pooling stages. Each one halves the grid, from 32 down to 2, so there is less work to do and each spot covers more of the image.
Deep networks shrink the grid deliberately. Fewer positions means less computation, and each remaining position covers more of the original image.
