Object Detection - Face Recognition

From the face recogniser’s tutorial. It seems it can only be used to recognise a face in a photo of a face (where the photo only contains a face that is a large part of it).

Also I see it can be used to recognise other things apart from faces.

So, can this not, alone, be used to do general object recognition in an image (where there are other objects) (would it have to be combined with Haar cascade object detection, first?)

true. it expects closely cropped face (or whatever) images. the less bg, the better.

also true. again, you have to detect and tightly crop the object, before you can run it through the prediction (same for the training)
(this is also true for most other “shallow” ml approaches, like ANN or SVM)