Convolution Output Dimensions | Dataford Interview Questions

Given a 500 by 500 by 3 image and 100 filters of size 5 by 5, what would the output be?

Assume each filter spans all three input channels, uses stride 1, and applies no padding. Implement a function that returns the output shape as (height, width, channels). For these inputs, the output is (496, 496, 100).

Input and Output

The input consists of an image shape, a square filter size, and a filter count. Return the valid-convolution output shape.

Constraints

image_shape contains exactly three positive integers: height, width, and channels
filter_size is a positive integer
filter_size <= height and filter_size <= width
num_filters is a positive integer
The convolution uses stride 1 and no padding
Every filter spans all input channels

Function Signature

def convolution_output_shape(image_shape, filter_size, num_filters):

Your question is Convolution Output Dimensions. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem