#0024

CNN Output Volume Shape

MediumML18 min15 XP

Problem

Given input height, width, input channels, filter count, kernel size, padding, and stride, return (output_height, output_width, output_channels).

Why This Matters

Model summaries are easier to debug when you can predict tensor shapes before running code.

Function Signature

def conv_output_volume(h, w, in_channels, filters, kernel, padding, stride):

Examples

Example 1

Input32x32x3 image, 16 filters, 3x3 kernel, padding 1, stride 1

Output(32, 32, 16)

Spatial size stays the same, but output channels equal the number of filters.

Constraints

Use the same kernel, padding, and stride for height and width.
Output channel count equals number of filters.
Input channels affect parameter count, not output channel count.

CodePython

def conv_output_volume(h, w, in_channels, filters, kernel, padding, stride):
    """
    Input example: 32x32x3 image, 16 filters, 3x3 kernel, padding 1, stride 1
    Return: (32, 32, 16)
    Return the value from the function so the tests can compare it. Do not print unless the prompt asks.
    """
    # write your code here
    return (h, w, in_channels)

def conv_output_volume(h, w, in_channels, filters, kernel, padding, stride):
    """
    Input example: 32x32x3 image, 16 filters, 3x3 kernel, padding 1, stride 1
    Return: (32, 32, 16)
    Return the value from the function so the tests can compare it. Do not print unless the prompt asks.
    """
    # write your code here
    return (h, w, in_channels)

Visible browser tests run here when available.

Testcases2 visible / 4 hidden categories

Same spatial size, more channels

Input32, 32, 3, 16, 3, 1, 1

Expected(32, 32, 16)

Sixteen filters produce sixteen output feature maps.

Stride shrinks spatial dimensions

Input28, 28, 1, 8, 5, 0, 2

Expected(12, 12, 8)

Height and width shrink, channel count becomes filter count.

Hidden Test Categories

Rectangular inputKernel size 1Padding 0Many filters