How to search for a specific thing in a model file using CoreML

I am trying to create an object recognition application where it can find a specific thing in a video frame.

As a beginner, I know how to let CoreML tell what it is on video using the rear camera of an iPhone. but how to reverse it by allowing the user to enter a specific string like people, dog, tree, etc., and create a bounding box around each one?

Thanks in advance for any suggestion