The simple act of observing the world without any kind of assistance makes us privileged people, but technology never stops breaking down barriers for those who have partially or completely lost their sight. One of the latest examples is Seeing AI, an artificial intelligence application developed by Microsoft that combines advanced APIs with verbal descriptions and optical recognition, whose goal is to "see" for those who cannot.
You visit the park, and a woman throws a frisbee. You go to a restaurant and choose what you want to eat from the menu. Then you have a presentation with your colleagues and dive headfirst into your daily work session. Now imagine doing all of that... while being blind. Think about the limitations, the difficulties. The world is not designed for blind people, and although there is always a possibility of adapting, it is never enough. In the past we have explored all kinds of advances, from bionic eyes to stem cells, but these are solutions trapped between "very experimental" and "too expensive". Is there a middle ground? Perhaps not in restoring sight, but there is something that can see for others, or at least, that is its main goal.
The application is named Seeing AI, a development by Saqib Shaikh and his colleagues in Microsoft's "Cognitive Services" division. Saqib went blind at seven years old, and since then he has known different accessibility options on computers, but he always had the idea of a system capable of describing what was happening around him, at any moment. The Seeing AI demo reminds us of two other Microsoft projects, How Old and emotion recognition, since besides recording actions and reading text, it can also describe the general characteristics of a person and their mood. Seeing AI is compatible with smartphones and Pivothead smart glasses.
Seeing AI had a relatively limited space at the opening of the BUILD 2016 conference, but without a doubt we want to see more about this platform. In fact, I believe that several of the aforementioned novelties stay in the background compared to the potential of something like Seeing AI...
