It is important that you provide a good training set that covers the types of variations you expect to occur in your testing set. For example, if you will only test with faces that are looking perfectly straight ahead (such as ID photos), then you only need to provide training images with faces that are looking perfectly straight ahead. But if the person might be looking to the left or up, then you should make sure the training set will also include faces of that person doing this, otherwise the face recognition algorithm will have trouble recognizing them, as their face will appear quite different. This also applies to other factors such as facial expression (for example, if the person is always smiling in the training set but not smiling in the testing set) or lighting direction (for example, a strong light is to the left-hand side in the training set but to the right-hand side in the testing set), then the face recognition algorithm will have difficulty recognizing them. The face preprocessing steps that we just saw
will help reduce these issues, but it certainly won't remove these factors, particularly
the direction in which the face is looking, as it has a large effect on the position of all
elements in the face.