Medium

Conv2D Output Shape

Medium

~12 min

code completion

Convolutional Output Shape

Before implementing a convolution, you need to know the output dimensions.

For a 2D convolution with no padding and stride $s$ :

H_{out} = ⌊ \frac{H _{in} - H _{f}}{s} ⌋ + 1

W_{out} = ⌊ \frac{W _{in} - W _{f}}{s} ⌋ + 1

where $H_{f}$ , $W_{f}$ are the filter (kernel) dimensions.

Example: 7×7 input, 3×3 filter, stride 2 → $⌊(7 - 3) /2 ⌋ + 1 = 3$ , so output is 3×3.

Your task:

Implement conv2d_output_shape(input_h, input_w, filter_h, filter_w, stride) that returns a tuple (out_h, out_w).

Example Tests

7x7 input, 3x3 filter, stride 1 → 5x5

Input: {"stride":1,"input_h":7,"input_w":7,"filter_h":3,"filter_w":3}

Expected: [5,5]

7x7 input, 3x3 filter, stride 2 → 3x3

Input: {"stride":2,"input_h":7,"input_w":7,"filter_h":3,"filter_w":3}

Expected: [3,3]

Non-square input

Input: {"stride":1,"input_h":10,"input_w":6,"filter_h":3,"filter_w":3}

Expected: [8,4]

You can read the full problem statement above. Create a free account to run code in the browser, submit solutions, and track your progress.