Convolutions 6 CNN

If I have a 57 pixel image and I want to do 6 convolutions with max pool of 2, stride of 1, padding of 1, kernel of 3, it is possible to do this

Have you tried it out or where are you stuck? :slight_smile:

when I do 6 convolutions with the operation: (((w-f + 2p) / s) +1) / max pool; for an image of 57 I have 0 of value

In that case it won’t be working and you would either have to increase the spatial size or reduce the pooling.