Core Building Blocks of CNNs

Convolution and Filters

At the heart of CNNs lies the convolution operation.

What is Convolution?

Convolution combines two sets of information:

1. The input image (matrix of pixel values).
2. The filter (a smaller matrix, like 3×3).

The filter slides over the image, multiplying and summing values to produce a feature map — highlighting where certain patterns exist.

For example:

By applying multiple filters, CNNs learn hierarchical features:

• Early layers detect lines and curves.
• Middle layers detect shapes and textures.
• Deep layers detect complex patterns like faces or vehicles.

Example:

Image (3x3):      Filter:         Result:
1 2 1             0 1 0           8
0 1 0     *       1 -4 1   =     -4
2 1 2             0 1 0           8

In CNNs:

• The image is convolved with several filters.
• Each filter learns a different visual feature.