Detailed explanation of conv2d of pytorch

In a convolutional neural network ,stride,padding, Followed by tuples (int,int), The previous one indicates the height , The latter represents the width

padding Detailed explanation of  

pytorch frequently-used padding function - Slow thick product - Blog Garden

padding =1 Express , The left, right, up and down are filled 1,

padding=(1,3) Express , Input the image of convolutional neural network , Height fill 1, Width fill 3

stride explain

PyTorch Learning notes (9)——nn.Conv2d And one of them padding Strategy _sooner High blog -CSDN Blog _conv2d padding stride=1, Express , When convolution , Right and down 1 Step  

stride=(1,2) It means , When convolution , One step to the right , Two steps down . The result is a constant height , Half the width .