Understanding average (sum) pooling in keras

Question

I have a simple sum pooling implemented in keras tensorflow, using AveragePooling2D*N*N, so it creates a sum of the elements in pool with some shape, same