Facebook found its way into the lives of the user in a very short span of time. It is the most used app in the world with more than 81% penetration. Zuckerberg just wanted to create a different user experience with Facebook, by letting people have a look at the other users’ life. Since then people have been using this social media platform to find people around the world. Now with evolving concepts and technology, Facebook is taking its user experience to a level higher through its ‘Portal’ and ‘Portal+’, loaded with rich features launched in November 2018. These features with smart speakers enabling video calls are the first of its kind in enhancing video calling experience.
What FB aims with Portal is the literal searching of people around the world, complying to the spirit of being a true search engine of people. Now how does the portal do this for us?
Portal is armed with a moving 12-megapixel camera with a 140-degree field view. This can follow you around the room keeping a watch over you.
So now what drives Portal and how has Facebook enabled this user experience of a video chat to seem like a normal face to face conversation crossing boundaries. This AI technology calls for introspection.
Smart Cameras in Portal
What is the idea behind Facebook’s smart camera?. The idea of the camera travels a little beyond the static shots captured by hangouts and similar platforms, allowing it to be a little creative in the shot captures. Facebook has tried to incorporate a little thought into the process of identifying points that require wide shots or zooming in for close-ups. This is where AI comes to play applying the thought process of a movie director.
Giving this an extra edge Facebook worked with filmmakers, going into the depths of the techniques they employ and converting them into Machine learnable insights. Working on this Facebook engineers tried out different proportions with landscape and portrait shots creating models for Portal with insights for decision making that is similar to roping in a human touch of applying thought to action.
It reportedly took Facebook two years to build this AI driven camera that can finally enable a hands free calling experience that cleared the limitations of distance. It simulates the experiences according to the situations it depicts. The algorithms that work for a one to one conversation is different from that with conversations taking place in a packed atmosphere. Portal decides on which shot to zoom on and how to follow individuals who are out of frame according to insights. Similar is with the lighting effects and the movements involved. Portal judges these situations well to give the best output.
Actually, all this can be captured from a moving camera, but as there was a huge time lag involved, Portal makes use of a wide-angle lens where all the movements can be edited digitally. According to Eric Hwang, one of the engineers behind Portal, it is not the technology of capturing the people around, but the innovation of capturing the relevant people on the device that is achieved through a small chip. These types of AI tasks would usually require large servers. But portal achieves this through compressing complex computer vision models that can run accurately and reliably.
Portal is the only product of its kind in the market. It was designed to make it easier to connect with the people closest to us and it is part of Facebook’s mission to achieve this through Portal.
Currently faced with privacy issues Facebook, however, assures that it does not keep or track the contents of Portal video calls, as Portal’s AI is run locally on the device, and not on Facebook servers. It claims that users can delete their voice history in Facebook’s Activity Log at any time.
However, it will be early to decide on the privacy of data shared over this new technology, as there no getting around the fact on the degree of data collection taking place.