The image recognition and tracking, and target following part isn't too far off. But intelligent collision avoidance, and true hands-off failsafe is quite a long way off. So you'll have to use these things out in the open where nobody is around, or at altitudes way above the trees, lift lines, etc...
Plus, I wonder, at what point are people going to get sick of watching themselves? Or watching video of other people? Do we really need everybody and their dog having perfect aerial video of themselves doing mediocre tricks?