The main objective of convolution in relation to ConvNet is to extract features from the input image. This layer does most of the computation in a ConvNet. We will not go into the mathematical details of convolution here but will get an understanding of how it works over images.

The ReLU activation function is extremely useful in CNNs.