Conv Output Dimension

Problem

Compute the spatial output size of a convolution: floor((n + 2 * padding - dilation * (k - 1) - 1) / stride) + 1.

The signature is conv_output_dim(n, k, stride=1, padding=0, dilation=1). Return an integer.

Hint

Use integer division for the floor, and remember dilation widens the effective kernel.

case 1

in: [28,3]

expected: 26

case 2

in: [28,3,2,1]

expected: 14

case 3

in: [32,5,1,2]

expected: 32

case 4

in: [7,3,1,0,2]

expected: 3

case 5

in: [10,3,3,0]

expected: 3

Starter codePython

def conv_output_dim(n, k, stride=1, padding=0, dilation=1):
    # Your code here
    pass

5 test cases · run your code against them in the editor