Our face has several features that can be identified, like our eyes, mouth, nose, etc. When we use DLib algorithms to detect these features we actually get a map of points that surround each feature.