Difference Between Channels and Kernels in Deep Learning

First of all, let’s understand the difference between Channels and kernels in the Laymen language.

Suppose you are in your favorite restaurant, a waiter comes and gives you a menu card. You observe that the menu card is divided into two parts, Vegetarian and Non-vegetarian. Again vegetarian and Non-Vegetarian sides have different food menu options.

Now, In this case, your channels are Vegetarian and Non-Vegetarian sides, and your kernels are Food menu options for Vegetarians and Non-Vegetarians.

Now, again suppose in Vegetarian, you have a food menu option as a Salad and, it is again divided into different types of salads, then your channel is Salad, and kernels are different types of salads.

If we try to understand the difference between channels and kernels in technical language, we can say that channels are the information that we are looking for and kernels are the feature extractors or filters.

Suppose we have two images of a happy and sad man. In these two images, channels are his lips, eyes, eyebrows, etc. And kernels are the size of his lips if he is smiling then his lip size will be bigger, and if he is sad then his lip size is normal. Same for eyes and eyebrows. In this way, our algorithm will detect whether the man is sad or happy.