I'm new in deep learning world, I want to develop a project regarding live text detection on mobile camera (without taking an image by the camera), can I do it by the tesorflow object detection api ? or with anyother tensorflow api or with opencv, which way is best? other suggestions are also welcome.