Nah it's just the latest hype to deep learn everything with lots of variations. It'll hit it a local maxima soon of what's achievable and then will begin the task of integrating actual computer vision knowledge.