0

I have an ML model where I get the convolution output as 11x11x64. Now I have to perform maxpool with pool size (2x2) and stride=1 with padding 'same'. I know the output size will also be 11x11x64. But what I don't understand is how the zeros padding is going to take place around the image of 11x11. I have trouble visualizing it and understanding how many zeros to put around the image. Please, someone, explain.

Utkarsh Kathuria
  • 1
  • This is explained in the answers to [What's the purpose of padding with Maxpooling?](https://datascience.stackexchange.com/q/67334/135707). Also see answers to [What is the difference between 'SAME' and 'VALID' padding in tf.nn.max_pool of tensorflow?](https://stackoverflow.com/q/37674306) on Stack Overflow. – Lynn Oct 16 '22 at 07:06

0 Answers0