An introduction to cascade classifiers

What are cascade classifiers? Let's take a look at both the words individually and then combine them to see what the phrase actually means. Classifiers are like black boxes that classify objects into various classes on the basis of a training set. Initially, we take a large set of training data, feed it to any learning algorithm, and compute a trained model (classifier), which is capable of classifying new unknown data.

Let's understand the word cascade. In the literal sense of the word, cascading means to form a chain. In the current context, cascading implies forming a multistage classifier, where the output of one stage is passed on to the next stage, and so on. Cascade classifiers are used in situations where you have low computational power and you do not want to compromise on the speed of your algorithm.

Cascade classifiers that will be covered in this chapter are as follows:

• Haar cascades (Viola and Jones – face detection)

• LBP cascades

Let's briefly...