2d convolution for stride of one with same output shape

2d_convolution.py

import numpy as np

def convolve2D_same_output_shape(input_volume, kernel):

    kernel = np.flipud(np.fliplr(kernel))

    kernel_height = kernel.shape[0]
    kernel_width = kernel.shape[1]

    input_height = input_volume.shape[0]
    input_width = input_volume.shape[1]

    padding_height = int((kernel_height - 1) / 2)
    padding_width = int((kernel_width - 1) / 2)

    output_volume_height = (input_height - kernel_height + 2*padding_height) + 1
    output_volume_width = (input_width - kernel_width + 2*padding_width) + 1

    output = np.zeros((output_volume_height, output_volume_width))

    assert output.shape == input_volume.shape, f"Expected output.shape == {input_volume.shape}. Got {output.shape}."

    input_padded = np.zeros((
        input_height + 2*padding_height,
        input_width + 2*padding_width
    ))

    input_padded[
        int(padding_height):int(-1 * padding_height), 
        int(padding_width): int(-1*padding_width)
    ] = input_volume 

    for x in range(input_height):
        for y in range(input_width):
            output[x, y] = (kernel * input_padded[x: x + kernel_height, y:y + kernel_width]).sum()

    return output

if __name__ == "__main__":

    kernel = np.array(
        [
            [-1, -1, -1], 
            [-1,  8, -1], 
            [-1, -1, -1]
        ]
        )
    image = np.random.randint(0, 255, size=(128, 128))
    output = convolve2D_same_output_shape(image, kernel)